Executive Summary and Summary Report of the Steering Committee Computing for Experiments CERN Data Networking Requirements in the Nineties Accelerator Computing Requirements in the Nineties Computing for Engineering at CERN in the 1990s MIS Computing in the 90s Theory Computing in the 1990s
July 1989
This Report of the Steering Committee sets out not to make a précis of the content of the working group reports, which already contain summaries of their own, but rather to put the requirements into an overall picture. It indicates the key directions that emerge, and the consequences they have on resources both at CERN and at the level of European High Energy Physics, (HEP).
Part 1 of the report contains an Executive Summary of the recommendations of the Steering Committee.
Part 2 reviews the existing computing facilities, and overall trends and directions, distinguishing between the computing requirements of the CERN laboratory infrastructure and the needs of the experimental programme. Estimates are given of the total expenditure on computing within the laboratory over the past decade, and of the computing power available to HEP elsewhere in Europe.
Part 3 reviews the six working group reports and indicates the key directions and challenges. It makes recommendations to the managements concerned on resources, the evolution of activities in certain areas and next steps in the planning.
Finally, three appendices cover:--
This Executive Summary of the Recommendations of the Steering Committee necessarily abbreviates the arguments and recommendations of the Summary Report. In particular, 17 recommendations of the Summary Report have been compressed into one major recommendation 5, "Major Components of the Computing Plan", which cover the whole of section 1.3 below. For this reason, cross--references are given to the relevant sections of the Summary Report and to the Summary Report Recommendations (SRR).
The Steering Committee endorses the recommendations of the Working Group reports within their context as requirements documents. It recognizes the strong level of agreement and support from the user community for the requirements. However, the total resources required for full implementation of these recommendations, in particular those of the physics programme, are extremely large relative to current expenditure and staffing levels, and in some areas are a factor of two to five higher.
The Steering Committee is aware that there is a widespread shortage of money across the whole laboratory, and a commitment to reduce the overall staff complement. It understands that it is unrealistic, in the present CERN budgetary situation, to expect that all of the requested resources could be provided on the timescales mentioned. Nevertheless, it believes that it is essential to find a means to increase the resources allocated to computing in order to carry out some reasonable fraction of the recommendations.
It is strongly recommended that the resources allocated to computing be increased above the present level, and that this be reviewed in the context of the overall CERN scientific programme. (SRR 1, section 3.2.3) |
The organization required for collaboration on computing matters at the European level needs a fresh approach. The need for easy movement of programs and data between home institutes and accelerator laboratories, for improved facilities when working remotely, and to make good use of all available resources, makes improved collaboration vital.
The present HEP--CCC should evolve into a fully representative body with a strong mandate to coordinate European HEP computing. In order to help accomplish this, a set of advisory groups which carry out the technical coordination in the different areas of HEP computing concerned is recommended. (SRR 3, section 3.2.4) |
We strongly support CERN's long--standing policy that, as a guideline, two--thirds of HEP computing resources related to the analysis of data from CERN experiments should be located in the home institutes. However, until now, application of this policy has been limited to pressure on the individual experiments and post--facto reporting. The demand for processing resources, especially from the LEP experiments, will be so considerable that an improved integration and planning on behalf of the experiments and the computing centres providing the resources is essential
A new method is needed of organizing and planning the allocation of mainframe production resources amongst the larger European centres, for the major CERN experiments. This should seek to minimize the barriers to moving production work, for instance through improved telecommunications facilities and standardized working environments. (SRR 2, section 3.2.4) |
The presence of large private computers on the CERN site goes against the spirit of the one--third:two--third policy above. In practice it also places an additional load on CERN resources. The committee recognizes the difficulty of making hard rules in such cases, and of refusing to accept the installation of private resources at a time when CERN's general resources are scarce.
The setting up or expansion of "private" computing facilities at CERN should be carefully controlled. In addition, the allocation of central facilities to an experiment should take into account the availability of "private" facilities. (SRR 15, section 3.2.11) |
We recommend that the plan for computing in the next five years must provide for the following key elements:-- |
Parallel computing systems have in addition considerable interest for on--line systems and applications for theoretical physics.
The laboratory is seriously under--staffed for the computing demands it is already asked to fulfil. The working groups make large additional demands for staff. (On face value, they ask for around 100 posts). Furthermore, in view of the overall compression of the CERN staff complement that is in progress, it will be particularly difficult to make the necessary increase in informatics staff numbers in the short term. The committee therefore feels that:--
It is essential to agree on a staged plan for an acceptable increase in informatics staffing levels across the whole laboratory over, say, five years. (SRR 25, section 3.8.1) |
A small, but valuable, contribution to alleviating the staffing problem is being made through students, coopérants, joint projects with industry, etc.. Some increased activity of this nature may be possible, although it is clear that the essential part of the problem cannot be solved by this means.
We recommend investigation of further alternative methods for informatics staffing,
such as the appointment of technical associates from companies and universities, and
increased involvement of graduate students. Joint development projects and staff exchanges with industry and academia, especially in the fields of advanced electronics and computer science, should be actively encouraged. (SRR 26, section 3.8.1) |
Despite the urgent pressures from the above situation, a massive effort is necessary in order to train and re--train existing CERN staff in informatics. If the missing computer--literate staff are to be found, a substantial part will have to come by internal movement and re--training. This will require a training programme on a scale comparable with that found in industrial companies which have successfully tackled this problem before CERN. New attitudes and extra resources will be required to produce the desired results.
CERN must set--up an organization devoted to a massive training and re--training of the existing staff in informatics technology. (SRR 27, section 3.8.2) |
In an associated area of training:--
Progress in the field of computing is so rapid that CERN must devote a fraction of its staff resources to work on advanced development projects that are only expected to show benefits in the medium to long--term. It is also vital that the long--term CERN staff involved with computing are kept well informed of the latest techniques. (SRR 28, section 3.8.2) |
|
In making the plans for "Computing at CERN in the 1990s", it is necessary to consider two major components. On the one hand, there is a responsibility to provide a healthy and broad computing infrastructure for the 3500 or so staff, plus the users based in the laboratory. On the other hand, there is a special responsibility to provide for the specific needs of the experimental physics programme which involves over 4000 visiting scientists, coming from some 150 institutes in Western Europe, and also from further afield.
It is well understood that this report comes at a difficult time for several reasons. The physics programme of the next few years will be dominated by the LEP experiments whose computing needs are very demanding in resources. The laboratory computing infrastructure is still in a build--up phase. The level of agreement on the requirements in both these areas within the user community has been remarkably good. Yet these requirements come at a time when an overall policy to reduce CERN staff, leading to minimal recruitment, coincides with a period of unprecedented budgetary difficulties for the Laboratory.
Planning for CERN cannot be made in isolation from planning for the whole European High Energy Physics community. The challenge posed by LEP is such that the level of coordination and cooperation at a European level will have to increase significantly, especially if the above budgetary difficulties are to be overcome. Explicitly, the required computing resources at CERN will have to be more than matched by increased resources in the collaborating institutes, and specific steps, particularly in the area of telecommunications, will be required to facilitate the exploitation of these distributed resources.
In a planning report of this nature, it is necessary first to establish the "base--line" on which the proposed changes can be built. This chapter contains a brief review of the computing facilities already in place at CERN at the beginning of 1989, together with the staffing levels and the money spent on computing across the laboratory, over the past few years. For more than a decade there has been an exponential growth of the installed computing power, a strong growth in the range of facilities and in numbers of users served, whilst the staff numbers and budget have remained essentially constant.
Amongst the initiatives that have made this possible is the establishment of exceptional relationships with major manufacturers. These have led to outstanding financial conditions and to joint projects that have opened doors to advanced technology and industry collaboration. CERN has something to offer in return provided it is prepared to keep its computing environment at the forefront of technology. On the other hand, if CERN is unable to maintain a leading "shop window" position, it will lose the interest of these industry collaborations, the important discounts and the advanced technology that it needs.
The conscious provision for the CERN general computing infrastructure is relatively new. Ten years ago, such general facilities were provided only as a by--product of those installed for physics number--crunching or data acquisition. Their suitability varied with the task in hand. WYLBUR provided for simple general needs, but could not prevent isolated initiatives, leading to inefficiency typified by the existence of eleven different makes of word processor on--site. This being recognized, some actions have been taken:
There are similarities between these three infrastructure areas. Firstly, the planning for all three has been reviewed and agreed relatively recently by the CERN management. Consequently, few major changes are revealed in this report. Secondly, working organizational structures for overseeing the activities are in place. Thirdly, the overriding policy of reducing CERN numbers hits hardest activities which are being built--up. Paradoxically, the recognition of manpower wasted in isolated efforts leads to the need for organized support, for which the staff are difficult to find. In this situation of few new posts, it seems clear that the major part of these support posts can only come from reorganization and re--training within the laboratory itself.
Computing support for the physics programme has changed significantly in the past five years. The range of machines, systems and facilities has had to expand dramatically, along with the physicists' requirements. The complexity and degree of interconnection has obliged the support staff to become highly specialized, and some have moved from this traditional area into the newer areas above.
Faced with this situation, many solutions have been tried; stopping services, pushing more on to the users, greater standardization, use of commercial packages etc. Solutions at a European level, albeit needing greater organization than in the past, can no longer be ignored.
The next five years of the physics programme will be dominated by LEP. The high statistics required will lead, both at CERN and in the home institutes, to high demands on processor power, as well as on data storage and handling, all of which will be major financial investments. Just as the manpower available to HEP at the European level will have to be pooled and organized better than in the past, so will these major computing resources, e.g. the processor time and data handling facilities available for physics production in the major European centres. Both will be scarce.
The rest of this chapter records the existing computing resources, staffing and costs as a "base--line" for the discussions of the working group reports that follow. A summary review of the computing facilities presently installed across the laboratory can be found in Appendix A. Some overall considerations are discussed here.
A summary of the computer power currently installed in the CERN Computer Centre is given in [Ref.]. As a comparison, Appendix C contains an estimate of the corresponding power available to HEP for data processing in the outside European centres. Note that this estimate attempts to make the difficult distinction between resources organized for major physics production computing and the smaller machines and workstations that are more readily used for physics analysis. Taking into account the fraction of the CERN capacity that is available for physics production, say 70 to 75 units, the potential capacity in member state institutes during 1989 matches reasonably well the "one third/two thirds rule". Some further capacity is available to the LEP experiments outside the member states, and in "private computing facilities" at CERN.
The number of users of the central computer services is high. As an example,
The above examples all refer to the computer power and range of services within the traditional computer centre. These are perhaps the easiest to quantify. Yet it is above all in the past five years or so that an explosion of many other forms of computer services across the CERN site has taken place. As an example, networking, both on and off the site, has seen dramatic growth, and the services thus made available have become a normal and fundamental requirement of every computer user whatever his or her level of skill. Standardization on one form of network has been far from possible.
A second example, namely the growth in numbers of intelligent workstations, (from Macs/PCs to high--end Apollos/VAXstations), which have been installed on CERN office desks, is a witness of the spread of computers throughout the laboratory, and their use in an ever increasing range of activities. There are certainly more than 3000 such workstations installed in offices of staff and visitors on--site, and the rate of installation is increasing. Taking as an example the pattern observed in the USA, one can expect this number to exceed one per person, and hence to grow beyond 6000.
Without attempting to quantify further all areas of computing growth at CERN, the point that needs to be recorded as a base--line in this report is that right across the laboratory there is a general shortage of computer--literate staff, be they for direct support or for applications support, which has been estimated on several occasions to be between 80 and 100 persons.
Material costs labelled Central Computers in the CERN accounts
Note that the CERN accounts contain no explicit record of the "costs of computing" as such. Indeed, the term "costs of computing" needs definition in such a discussion. Three examples in common usage are, the costs of the computer centre, the full costs of the DD Division, or the costs coded in the CERN accounts under headings labelled "computers" for all divisions.
This rather flat expenditure profile across the years is also typical of the numbers
found in Appendix B for the computing costs across the laboratory as a whole. In order to
make this comparison for all divisions in a consistent manner over many years, one is
obliged to take a strict definition from the accounts, (see Appendix for this definition).
The recorded annual material costs for computing for the whole laboratory thus obtained,
have remained (in 1988 prices) between 27 and 37 MSF with average 31 MSF over the past
decade (see and
The personnel costs for computing are harder to determine. Whilst they are well defined and known for DD Division, their definition becomes hazier in other divisions. For example, the distinction between physicist or engineer and computer support staff can be difficult. Similarly, it is not easy to understand what fraction of the accelerator control staff should be considered as computer specialists. Nonetheless, an estimate is made in the Appendix of these personnel costs, using a method again chosen to allow a comparison across the years. This suggests that the personnel costs for computing have changed little over the years. Whilst the accounts show that personnel costs of the DD Division have in real terms stayed remarkably constant, this is unlikely to be true for the rest of CERN where the numbers of people involved in computing activities have grown substantially.
Using the above definitions of material and personnel costs, one may sum the total
costs attributed to computing across the years, (
This part of the report reviews the reports of the six working groups, attempting to integrate them into a consistent picture. It gives an overview of the key directions to be followed, and the consequences on resources both at CERN and at the level of European High Energy Physics. The position and attitude of the Steering Committee in making such choices and judgements is summarized in the following points.
The Recommendations that appear in boxes in the review are those of the Steering Committee.
"Computing for Experiments" is the largest of the working group reports and by definition the one most concerned with the physics programme. The report is the product of a number of smaller working groups whose recommendations were grouped into two main areas, separating data acquisition and all other aspects of the data processing. Part 2 of the Computing for Experiments report contains 22 recommendations, themselves a summary of a larger number of recommendations in the main body. Part 3 reviews costs and manpower whilst making further remarks on priorities. An Executive Summary of Computing for Experiments is provided in Part 1.
It is evident that detailed remarks on all aspects of this 120 page report cannot be made here, nor is it attempted to make a further précis. The major directions and challenges to be faced by the community are identified, and input to the next step of the planning is given.
The picture emerging for the years ahead is rather consistent and well agreed upon both by physicists who use computers and computing specialists. The key elements are summarized as follows:--
The consequences of the above directions are reviewed in the next sections.
The user requirements have been clearly analysed and presented in the working group report. They have received substantial support in public user meetings and in general discussion in the community. They present the Steering Committee and the managements of the various high energy physics institutes and funding bodies with a number of serious problems, not least a strong opinion that the present levels of funding and staffing are at a marginal or inadequate level. The costs and manpower suggested represent significant increases over present levels, (see Tables 1 and 2 of the WG report, deliberately NOT summarized out of context here). Whilst the above tables concern mainly CERN, there are clearly described consequences for the community outside. Apart from substantial increases in processing power and disks, there are high--speed line and workstation costs to be borne by the institutes or the experiments.
It is in this report above all that the greatest proposed resource increases from the present levels are to be found, especially when the consequences on the outside institutes are included.
Any study of requirements can make estimates for the additional costing and manpower of new activities. The difficulty of such a study comes in relating those requests to the current budgets and staffing; to fold in what can be achieved by stopping activities, redeployment, imaginative negotiation, joint projects with industry, spreading the plan over more years etc. Indeed, this is exactly the task facing the laboratory.
The magnitude of this task is considerable. The working group costs their requirements at 128 MSF spread over three years, including 26 MSF estimated for high speed links to be payed by the outside institutes. At face value, the resultant CERN cost of 102 MSF is a factor three to four higher than the DD capital expenditure which has averaged 8.6 MSF/annum over the past 9 years, and 10/11 MSF/annum in the past two. This money has to be found in addition to the 5 MSF outstanding payment for the most recent IBM 3090 upgrade which is already spread until 1992.
Similar remarks apply to the staffing estimates of this working group. They suggest that for the manpower resources "located at CERN and serving the whole of the CERN--connected community" an increase of new staff totalling 34 in 1989, plus 20 in 1990, plus 17 in 1991 should be made. Given the probable level of new recruitment that will be available to the laboratory as a whole, it is clear that if these numbers are to be in any way approached they will have to come in major part through diversion of staff currently employed on other CERN activities, including physics or physics support. Furthermore, all other creative forms of obtaining manpower for these activities will have to be explored, such as industry involvement, European collaboration amongst HEP institutes, and closer involvement of computing expertise in the universities.
The conclusion is inescapable that the resources required to fulfil the programme foreseen by this working group represent, even after drastic cuts, a significant change from the previous decade. The principle reasons for this change are to be found in the growth in data rates and event sizes at the LEP collider, and also in the advances in computing technology, which have opened up new possibilities and allowed greater sophistication in data acquisition and detector systems.
The working group, in recognizing this increased demand, states:-- ""computing expenditures provide at least equally significant potential in added physics returns when compared to spending for other sectors such as accelerators or detector construction.""
It is strongly recommended that the resources allocated to computing at CERN be increased above the present level, and that this be reviewed in the context of the overall CERN scientific programme. |
Many recommendations made in the working group's report concern the needs of the HEP community as opposed to specific needs at CERN. The CERN management has traditionally seen its rôle as that of the host laboratory, and has been cautious in addressing computing problems at the community level. The 'High--energy Physics Computing Coordination Committee' (HEP--CCC) exists as an independent, high--level representative body for community computing matters, and a number of community--wide technical groups exist, for example ECFA SG5 [Footnote: ECFA SG5 has recently evolved into the HEPnet Requirements Committee (HRC), partner to the HEPnet Technical Committee (HTC).] and HEPVM. The working group expresses a strong view that the time has come for a more concerted effort in this area. This growing demand is recognized, and the view of the working group endorsed. Indeed, improved coordinated planning at a European level is a major theme of "Computing at CERN in the 90s".
This leads to the following two recommendations:--
CERN's long--standing policy that, as a guideline, two--thirds of HEP computing resources related to CERN experiments should be located in the home institutes is strongly endorsed. It is understood that, until now, application of this policy has been limited to pressure on the individual experiments and post--facto reporting. This has led to periods where production resources available to HEP or provided specifically for HEP in outside computer centres have been under--utilized. On the other hand, note is taken of the strong view in parts of the physics community that the movement of physics production from CERN to outside institutes has traditionally introduced significant delays which are considered unacceptable at certain times, e.g. at the start--up of a new experiment when most of the key personnel involved are unavoidably at CERN.
The obstacles preventing the movement of production can be reduced. The provision of high--speed links, described in this chapter and the next, would make a significant jump in the convenience of remote working and remove many of the barriers and delays. The success of the HEPVM collaboration in an important fraction of the sites involved has already demonstrated that the time--wasting and frustrating differences between sites can be significantly reduced. There is now good evidence that attention to these convenience factors does indeed reduce the dependence on the CERN computer centre.
In the light of the very significant production resources, (e.g. CPU power, disk space, cartridge units and robots), being requested for the physics programme ahead, and the severe difficulties in funding those resources, an improved effort must go into making the best possible use of such production resources as can be made available at the European level. This will require a new coordination and planning on behalf of the experiments and the computing centres providing the resources.
A new method is needed of organizing and planning the allocation of mainframe production resources amongst the larger European centres, for the major CERN experiments. This should seek to minimize the barriers to moving production work, for instance through improved telecommunications facilities and standardized working environments. |
The initiative for setting this up should come from CERN, but it requires full support at the European level, and should involve those responsible for the management and allocation of resources in the major computer centres.
The present HEP--CCC should evolve into a fully representative body with a strong mandate to coordinate European HEP computing. In order to help accomplish this, a set of advisory groups which carry out the technical coordination in the different areas of HEP computing concerned is recommended. |
In reality this is not as trivial as it perhaps sounds, as the present HEP--CCC is itself aware. The national representative physicists on the committee often have only indirect control over the resources and running of the computer centres available to HEP. Where university centres are concerned, the special needs of HEP can conflict with overriding university needs. Coordination of working environments, software and hardware choices for HEP conflict with other bodies trying to achieve coordination in a different manner at say a national level. The size of the committee would be a problem if all members states and funding bodies were to be directly represented. These problems are being addressed already by the HEP--CCC. Part of the solution can come through the establishment of further strong technical groups reporting to the HEP--CCC.
The main elements in the evolution of centrally organized computing, at CERN, in national and regional centres, and in the universities, are:
It is important that this evolution take place smoothly. While we must press ahead with the introduction of new technologies for distributed computing, we must also learn how to exploit these effectively, and be careful not to abandon prematurely existing proven services."
This is clearly and widely agreed as the model for the years ahead for "physics off--line computing". It identifies a number of components and areas for work, notably:--
The recommendations as to specific areas of the above model are to be found in the following pages.
Data Storage and Handling are also part of the above model, but are treated separately because of the particularly important rôle they have to play. The physics to be done at LEP involves high statistics. Modern detectors produce event sizes measured in hundreds of kilobytes. The high requirements for LEP data storage were first identified in the MUSCLE [Footnote: The Computing Needs of the LEP Experiments, CERN/DD/88/1, January 1988] report. One number to remember is that each LEP experiment is expected to have an accumulated data volume of approximately 7000 Gigabytes by "1991" [Footnote: The MUSCLE Report defined "1991" as being the time when the total number of Z events that each LEP experiment has accumulated reaches 10 million. In practice this is not now expected to occur before the end of 1992.] . The working group, in section 9.2.4, argues that the MUSCLE numbers are lower limits, particularly with respect to "master DSTs", and adds to them the predictions of non--LEP experiments. They consider the distribution of data to universities and regional centres and the copies that implies. They suggest 300,000 active tape cartridges [Footnote: at the current 200 MBytes capacity] in 1992 at CERN and at least that number outside CERN. This leads logically to a proposal for automated handling of cartridges, with a capacity of 40,000 cartridges in 1991 at CERN, with similar robots in major centres outside CERN. They propose a centrally supported cartridge/tape copying and distribution service such that cartridges can arrive at regional centres and universities within days of submission of an appropriately authorized electronic request. Regional centres should have similar facilities.
Following the MUSCLE report, they argue that 100 GBytes of disk space for active data per LEP experiment by end--1991, representing 1.4% of the expected data volume, are necessary in order to facilitate physics analysis and eliminate a large fraction of the cartridge mounts. In this they follow the MUSCLE model of a central hierarchical data repository, from which subsets of data may be extracted for different physics analysis needs elsewhere on workstations or off--site. They plausibly suggest that the disk storage for all of CERN should be 8 to 10 times bigger than one LEP experiment. Adding system needs, this leads to their requirement of 1000 GBytes of disk storage by 1991.
Tools for managing the data in the above implied storage hierarchy are another requirement, and should include automatic migration of data based on activity.
The cost of the disk space requested at CERN is realistically estimated by the working group to be 31 MSF, including the required controllers etc. Given this large sum, which represents about 30% of the total estimated costs of this working group, (excluding the long distance lines), three obvious questions are:
The following recommendation is suggested as a reasonable method to deal with this situation. It would allow some form of optimized planning which could approach the desired capacity at one third of the working group's proposed cost. It will allow some relief to a situation that will become very difficult by the end of 1989. Should the capacity which it is possible to install with this reduced budget prove largely insufficient in later years, not only will the costs per byte have fallen but the budgetary situation should have improved.
A fixed budget of the order of 2.5 MSF/annum over each of the next 4 years should be allocated at CERN for this important acquisition, such that the planning and the disk space eventually obtained can be optimized. |
This is clearly a crucial matter for the detailed planning that follows.
The automated cartridge handling device is an important proposal which no longer requires a recommendation in this report in that the matter is already in hand. An attractive joint study agreement with a manufacturer to develop such a robot to CERN's specific needs has recently been signed, following Finance Committee approval. If successful this should fulfil the needs of the working group for a cost around 40% of their estimated 4 MSF. The funding for the robot must, nonetheless, be included in the budget planning. Note that it is possible that future announcements of higher capacity cartridges, may require money above that presently foreseen.
The exact levels of money and manpower required for a cartridge copying and distribution service are clearly dependent on scale and the exact requirements of the physics groups. This needs further investigation before it can be correctly costed. The working group's estimate of 0.5 MSF per year should in principle be sufficient to cover additional contract manpower, new hardware and maintenance money for a new service. The first step is to use existing equipment until it proves inadequate. One must not forget however that this is a service on which the LEP experiments are counting, and one which is completely consistent with ideas to make as best use as possible of the computing resources outside CERN.
The growing user pressure to provide for video 8 Exabyte technology as a second standard, in addition to the established 3480 cartridges (which now represent 85% of the mounts in the computer centre) is difficult to resist. However, the consequences on the costs and staffing of a centrally organized service using this technology could be considerable.
A centrally supported cartridge/tape copying and distribution service should be provided at CERN. |
It is widely believed that conventional mainframe computers will continue to be vital components of the HEP computing environment. The transition from terminal to workstations will certainly allow more productive physics analysis, but it is very unlikely to reduce the steady growth on demand for mainframe power, (in fact the contrary is likely).
The model in which the central computer is the master file server and base for cooperative processing with the workstations leads to high bandwidth demands for I/O which will require significant CPU power to drive it. Present measurements indicate that an aggregate throughput from the central data store to the workstations of 1 MB/s would require more than 50% of one of the IBM central processors.
The rôle of the central mainframe, apart from traditional general--purpose computing and part of physics production with its unavoidable development and debugging, will thus be as the central data repository and data handling system, exploiting a storage hierarchy based on the cartridge robot and the above disk storage, and feeding the workstations and parallel farms described below.
The CERN physics programme will require at least 600 CERN units of CPU power by 1991--1992. It is proposed that at least 150 such units be installed in the form of conventional mainframe systems on the CERN site and that hence at least 300 units be installed and accessible to HEP in the outside centres. The balance should be provided by parallel farms, as discussed below. |
By comparison the power installed in the CERN Computer Centre at the beginning of 1989
was 88.5 units, and that installed in other European centres available to HEP and equipped
for production was around 150 units. (See [Ref.] and Appendix C).
The recommendation is thus to continue to double the mainframe capacity in the next 3.5
years, following the line of the graph in
There is a conviction within HEP that the inherent parallel nature of the event--related computing problems should be exploited. It is proposed that the balance of the CPU power requested above be provided in the form of parallel computing facilities, which must be integrated into the normal operation of the large centres concerned, and which can then be used for production work with relatively stable problems. On the other hand, there is no general agreement within the community as to the true cost--effectiveness of, and to how much manpower has in reality been consumed by, the various existing initiatives using parallel farms in an off--line environment up to now, (whatever their success within specific experiments or in the on--line area).
It is recommended that a new generation of parallel processor farms should be investigated, using as far as possible commercially--based products, by setting up a pilot project. The farms should be fully interfaced to the existing off--line environment and operational procedures. |
Bearing in mind the targeted operation, the investigations will have to take into account the costs and manpower required to build and to integrate the system into normal operation, as well as reliability, available de--bugging tools, overall efficiency, etc. An initial budget of 0.5MSF for the pilot project should be reviewed according to success.
Whilst such investigations should exploit the obvious "trivial" parallelism at the event level, they should also address what other forms of parallelism could be profitable. Note that such parallel devices clearly have other areas of similar and related interest such as on--line and theoretical physics applications.
The working group suggests that around 150 units of processing power be provided in this fashion by 1991/1992, with one third of this at CERN. The working group notes that if one is unable to provide around 150 units of this type in a cost--effective and sufficiently useful manner, then this power will have to be provided in the form of conventional general purpose computers.
At CERN the proposal is approximately to double the existing capacity by 1991--1992. The cost depends on which computers one changes, and when, and strongly on possible price negotiations. The working group has foreseen 47 MSF for this expansion over 3 years.
In order to make an estimate of what might be possible, it should be recorded that CERN has spent some 29 MSF since December 1984 in order to acquire a net increase of approximately 43 units of IBM--compatible CPU power, together with 150 GBytes of disk space, and other peripherals such as cartridge tapes and communications controllers. Assuming the mainframes themselves represented about 75% of the cost, one has spent 22 MSF on 43 computing units over 4 years. It must be emphasized that these were skillfully negotiated prices, far below the levels obtainable with normal academic discounts, and it may not be possible to match them in the future.
As a rough number, one may assume that the price of such mainframes falls by a factor 2 over 4 years, and hence one should be able to obtain a net increase of around 80 mainframe units with no peripherals for around 20 MSF spread over the 4 years, assuming a continuation of advantageous arrangements made with the manufacturers concerned. Of course, there will also be a need for peripherals other than the disks, which have been specified separately. This suggests that a sum of 5 to 6 MSF/annum each year for the four years 1990 to 1993 is the minimum that could provide the necessary capacity by the end of 1992 (which is the latest point suggested by the working group). The payments still to be made on the IBM 3090/600E until the year 1992 are in addition to the above calculation.
The above estimation is based exclusively on IBM--compatible computers. It should be noted that the Cray version would be quite different. The present Cray is on a lease/rental contract fixed at 3.6 MSF/annum. In mid--1990 CERN will have the option to change this for a Cray 2 at the same price, renegotiate for a Y--MP or a successor machine, or continue with the present X--MP. The choices will depend in part on the degree of success of use of vectorization by the physicists. Obviously, further models can be made, based on other manufacturers, or combinations of manufacturers.
A capital investment budget line of 6 MSF/annum for the four years 1989/1992 for the provision of upgrades to the conventional mainframe power is recommended. This should cover the mainframes and major peripherals except disks which are specified elsewhere. |
The evolution to a distributed computing environment will establish personal workstations as the principal tools for general--purpose interactive computing. The challenge will be to develop the support services, the network management and the tools to integrate workstations with each other and with central services for file management, batch processing, etc. Management of workstation system software, communications, and files will become a task of considerable size, probably comparable with the management of a mainframe computer centre. The working group recommends that both at CERN and in the regional centres these tasks be centrally managed for the benefit of the end users.
Centrally managed services should be established to provide workstation support services for a limited range of workstations on the CERN site. |
The central VAX Cluster acts as a focal point for the integration of the distributed VAXes on--site, including the VMS--based personal workstations that are an important component of the above section, as well as serving a large community of visitors who are VAX--based in their home institutes. In addition, they have been of great value for interactive program development and for applications that only run, or are best performed on a VAX. Experimentalists working with VAXes used for data acquisition have profited greatly from the service.
The Cluster has provided a minor part of CERN's batch capacity and should not be upgraded in order to provide batch processing power, unless there are substantial technical changes in the future which would make this advantageous. However, the rôle that the Cluster plays as a centre of competence and focal point for the management of distributed VAXes and in particular the VAX workstations will increase in importance as described above.
It is recommended that the central VAX Cluster continue to be upgraded by a reasonable level of investment, in order to act as a VAX competence centre and management base for the VAX workstations. |
Note that the possible migration from VMS to Ultrix is another parameter which could influence the development of the cluster.
The Computing for Experiments working group is of the opinion that, whilst current generations of experiments are relatively well understood in this area, a major challenge on the horizon is that of constructing trigger and data acquisition systems for high rate hadron colliders such as the LHC. They consider that a great deal of research, development and learning of techniques from outside HEP is necessary. This must be started as soon as possible. It leads to recommendations how to face this challenge, to set up the correct framework and to train the people concerned, summarized here as:--
It is recommended to set up coordinated pilot projects for introducing new high--level design methodologies for both software and hardware, and for acquiring familiarity with the application of modern techniques to future experiments' real--time problems. |
The working group believes that smaller experiments have needs in this area which are often as technically demanding as those of larger experiments. These needs arise both from complex triggers and from data acquisition systems containing mixtures of on--line computers, workstations, and microprocessors. Due to their size, the smaller experiments are less able to provide for themselves. They are thus interested in standard, off--the--shelf systems. The view that if CERN wishes to continue its current broad programme, it is important that these experiments are not neglected is fully endorsed.
Centralized support for hardware and software components to be used in the construction of on--line systems at the experiments, as well as design support to aid their effective integration, will still be necessary, both for smaller experiments, and to provide components for the larger experiments. |
Very sophisticated design skills are required in these areas; the application of VLSI techniques to trigger and data acquisition problems is now of crucial importance. Additionally, the standardization of real--time kernels and associated system software should be followed closely, and taken into account in planning future data acquisition systems.
The general desire to attack the diversity of software can be found in many of the working group reports. In the report for experiments the strongest recommendation in this area proposes "to start a Unix service on the central IBM service at CERN as soon as adequate software is available from IBM." An initial small--scale service should receive further resources according to user demand. In addition there are a number of recommendations aimed at improved organization and coordination of the HEP software community.
This is not the first time that a pilot Unix service on the IBM has been proposed. What is new is that there are now sufficient good reasons for such a pilot investigation to make sense. Apart from the penetration of Unix into the HEP world though its appearance on workstations, the Cray and various other machines, several of the traditional arguments against Unix in this rôle are decreasing in validity. One may note in this context the previously bad reputation of FORTRAN compilers available with Unix, which gave inefficient run time performance compared with those available in proprietary systems (e.g. MVS, SCOPE, VMS, VM), There has been no seriously supported version of Unix on IBM systems. The recent initiatives of IBM in the direction of the AIX operating system, (AIX will run the same FORTRAN compiler as VM/CMS and MVS), and the creation of the Open Systems Foundation [Footnote: There are currently nine sponsoring manufacturers and more than 90 full members.] by many manufacturers, show an increased interest in Unix from the computing world in general.
Nonetheless, one is obliged to take into account the special needs of the HEP
community, particularly in the areas of very substantial tape handling and of batch
schedulers. These are not yet guaranteed to be solved by AIX on the IBM, even if they have
indeed been solved on the Cray. Indeed the requirements for Unix running on such large
systems have been tackled in relatively few places as yet. However, referring back to
It is recommended that a pilot Unix service is started on the central IBM computers at CERN as soon as practicable. |
The recommendations concerning the organization and coordination of software issues in the HEP software community include the proposal for a Software Support Committee. Such a body could help to reduce the diversity of software by making successful commercial and HEP--specific products more widely available. It could also evaluate the benefits of writing HEP--specific tools against the cost of manpower, were a suitable commercial package not available.
HEP--CCC should establish a Software Support Committee as one of its technical working groups. |
The crucial rôle that data base techniques and access methods have to play in the management of the huge volumes of data expected in the next decade of computing at CERN is recognized. The working group recommends "the consolidation of the use of ORACLE in the HEP community for this purpose." This recommendation is strengthened by the request for HEP--wide coordination of the definition, writing and installation of data base orientated software for large physics experiments. It is felt that this coordination could also come under the umbrella of the Software Support Committee.
Additionally, the working group report highlights the success of some larger experiments in using software engineering tools to design and implement their physics codes. They recommend the support of software tools and methodologies [Footnote: A methodology is a set of "methods", typically purchased as a collection of CASE tools.] , including their selection and evaluation. It is believed that, in view of the distributed nature of the software development effort amongst collaborating institutes, such tools will become essential for coherent and correct code.
The presence of large private computers on the CERN site goes against the spirit of the one--third:two--third policy. In practice they also place an additional load on CERN resources, and there are a number of other negative remarks that are frequently made against them. The interconnection and networking of such centres with the CERN Computer Centre causes arguments of unfair assistance to one experiment over another, whereas artificial restrictions in this area often make technical nonsense and lend weight to the argument that the money made available for the private centre would have been better invested for the community as a whole via the central pot.
To these traditional arguments against private centres, one may now add the complication of deciding what exactly is a private centre. In the case of a major mainframe this is clear, but the potential processor power of a networked set of powerful workstations, or an emulator farm, may well be higher. The confusion is all the greater for emulator farms used both on--line during an experimental run and off--line outside runs. The difficulty of making hard rules in such cases is recognized.
A further aspect of this difficult subject concerns the possible inconsistency of refusing private resources, and hence increasing the general CERN demand for resources, at a time when CERN funding for those resources is too scarce to meet the essential requirements for computing resources on site. This is definitely true in the case of funding for private centres that would otherwise not be available to CERN, or at least to the experiments concerned.
Individual cases should be resolved at the Directorate level.
The setting up or expansion of "private" computing facilities at CERN should be carefully controlled. In addition, the allocation of central facilities to an experiment should take into account the availability of "private" facilities. |
The existing mechanisms and structures for networking are well established and working, such that the overall theme is one of continuity and expansion, and of reinforcement of the existing data communications strategy. This does not imply that there is little to be done. The successful implementation of the networking requirements below is of the greatest importance to HEP computing and to its evolution towards cooperative distribution of function. Two major thrusts are clear:
The working group report gives a general model of computing activity. It estimates the requirements up to 1993 and lists the services required. It also reviews the state of the requirements made in 1983. There is close agreement with the body of this work.
It notes that three of the requirements from 1983, namely:--
have not yet been implemented in their entirety due to the lack of suitable (100 MHz) industrial products for the backbone, but an intermediate backbone network has been created using slower products. G.703 equipment has been widely installed, mainly by SPS/ACC for the LEP site, and it is used both for the digital telephone and for the intermediate backbone.
The Committee considers that CERN is well placed to exploit the FDDI products that will appear imminently on the market in order to provide this general--purpose high--speed backbone, and recommends active continuation of the FDDI programme. |
The Committee endorses the working group's view on protocol requirements which may be summarized as, protocols should wherever possible be manufacturer independent and standardized, but ad hoc or proprietary protocols should be used when performance or timescales demand them.
This section contains a summary of the new requirements from the working group which are felt to be most important.
The working group recommends that ""access to CERN at speeds of at least 2 Mbits/s is required from major HEP sites in Europe, to allow decentralized analysis of the LEP data, as soon as this analysis starts in earnest. Higher speeds (equivalent to those attainable on a LAN) will be required as soon as practicable, to allow effective use of workstations over geographical distances.""
The provision of these high speed links is of the highest priority, and initiatives that are being actively pursued are strongly supported. It is recognized that the implementation of 2 Mbit/s links on CERN site, namely CERNET, enabled a significant change in the number of users able to work away from the Computer Centre. The recent success of T1 (1.5 Mbits/s) long distance links within NSFNET [Footnote: The network of the American National Science Foundation, currently upgrading some links to DS3 (45 Mbits/s). ] in the United States is additional evidence of the quantum jump in user productivity that can be achieved through such bandwidth. Progress towards the desired goal of the HEP community to overcome geographic barriers, and to facilitate decentralized physics data processing/analysis, is dependent on the existence of such links. The required bandwidth has been expressed as "one glass--fibre equivalent".
Improved communication facilities between CERN and the collaborating institutes must be provided. The major paths should operate at a bandwidth of at least 2 Mbits/s. |
The provision of these links goes beyond the cost of the lines themselves, itself a challenge at present PTT tariffs. Within the major centres non--negligible funding must also be found for infrastructure costs such as multiplexors and communications controllers. These costs could well amount to around 150 kSF for each end of such a high--speed line, assuming the likely model of multiple protocols on each line. CERN has a policy that the costs of such outside lines are payed by the outside institute itself. Up to now this has consisted mainly of PTT bills.
The borderline of payment for the various components necessary to install these important lines at CERN must be urgently clarified. |
The smaller national institutes will only profit if they too can obtain funding for high bandwidth into their national/regional centre.
The Committee endorses the associated recommendation of the working group, namely that:--
CERN should take an active role in HEPnet activities, without losing sight of the priorities of the CERN experimental programme. CERN should also, in conjunction with HEPnet, collaborate with general--purpose network activities (RARE, COSINE, EARN, EUNET) as long as this is beneficial to the HEP community.
At a second level of priority, support is given for the two working group recommendations which are both aimed at stopping or discouraging solutions requiring special, (or worse, home--made), hardware or software, namely:--
"
In this context, CERNET is being actively phased out. The old TITN networks for the PS and SPS control systems should also be phased out.
A summary of the resources estimated by the working group is given below. For an explanation of the financial model used and of the assumptions made the reader is referred to chapter 6 of the working group report. The financial model in [Ref.] contains a certain contingency after 1990 for unforeseen capital expenditure. Money for the STK digital exchange and on--line networks (controls, DAQ) have been excluded.
Summary of Financial Totals from the Networking Report | ||||||
1992 | 1993 | |||||
Capital Expenditure (MSF) | 3.5 | 3.5 | 4.5 | 3.5 | 3.5 | |
Operations Expenditure (MSF) | 5.5 | 5.5 | 5.5 | 5.5 | 5.5 | |
1. including 4.1 MSF paid by DA for telephone, telex, X.25, EARN, EUNET
Two major items are not included in the above table. The first is the cost of the high--speed external links, assumed here to be paid by the institutes which request them, following established CERN policy for conventional direct links. At normal PTT tariffs this would require an increase on the community's operations budget rising to 12 MSF in 1992. This is covered further in the Computing for Experiments report. The second item is the infrastructure in the computer centres at the end of the links, intelligent multiplexors and communications controllers.
The detailed arguments and staff movements leading to [Ref.] are not reproduced here. Note that staff for on--line networks, TV, audio and safety communications are specifically excluded, and that it is assumed that all departures will be compensated, with the exception of some telephonists.
These are considered and conservative figures for the staff required to run the services.
>Summary of Staffing levels from the Networking Report | ||||||
EP/EF | All Divs | |||||
Current Staffing Level | 46 | 1.5 | 6 | 0 | 1.5 | 55 |
Future Staffing Needs | 52 | 1.5 | 1 | 1 | 2 | 57.5 |
1. all DD groups, not only CS, and includes the telephone service.
The computing requirements of the accelerator divisions bring together four somewhat different areas. Firstly, there is the well--established accelerator controls area, where the accelerator divisions have long been self--sufficient providers of their own computing services, and the specialist computer staff have had little dependency on others. In the second area of accelerator design, the simulation work has been largely performed using the general Computer Centre services. The third area concerns the accelerator engineers whose requirements clearly overlap with those in the Computing for Engineers working group report. Finally, the required management information services infrastructure largely overlaps with the MIS working group report.
It is worth noting that the accelerator divisions have used database management systems in earnest for the accelerator planning and installation for many years. Indeed, one can make the general observation that those divisions have played an important rôle in pioneering areas of the computing infrastructure at CERN such as CAE, MIS and DBMS.
In overall terms, this is a well--established part of the computing programme of the laboratory, and the structures required to deal with it are mainly in place. The trend is thus towards a common model for accelerator computing in the future, thus reducing the diversity of solutions in the different divisions. The manpower needed is dependent on the success of this common model.
The major themes particular to the accelerator computing requirements are thus:--
The working group has developed a schematic view comprising three levels of LAN. The first layer is a conventional office LAN with workstations in every divisional office, and servers of various forms in the corridors. The second LAN is used for the Main Control network, and is interconnected to the office LAN and to the central computer services. The third layer, named regional LANs, replaces traditional equipment control highways, such as serial CAMAC. The regional LANs will be connected in a controlled fashion to the other LANs.
The computing infrastructure of the laboratory has been almost totally based on the needs of the experimental physicists, a situation which is necessarily changing. As this infrastructure broadens, there is a general trend for the accelerator divisions to stop the specialist services they have provided for themselves in the past, and to use the general laboratory services. Particular examples are a centralized accelerator database, central networking support, central computers and their file backup services.
This is not a trivial point. The accelerator divisions have some particular demands to make on the general services which must be taken into account. The provision of an 24--hour database service for all accelerator data, including controls, needs consideration and management, and probably additional resources. Whilst the Computer Centre has often scheduled its services around the needs of a specific experiment(s), it has never directly had to give special priority to the needs of the accelerator and its engineers during running or development periods.
The communications facilities of the accelerator divisions should be unified using TCP/IP protocols.
A single central database management system for all accelerators, preferably running on the same computers, is proposed. The accelerator divisions should capitalize on their early investment in ORACLE.
Other standards such as EUCLID and AUTOCAD are proposed in line with the recommendations of the Computing for Engineers working group report.
It is recommended that a much greater fraction of the network facilities should be provided by DD Division, including some particular demands of this working group. The substantial commitment of the accelerator divisions to IBM--compatible PCs requires good interconnection of these PCs and full network facilities, including coordination of choice of software, negotiation of site licences, distribution, and grouped ordering. This shift of support was not foreseen by the working group on networking, and should be referred to the Telecommunications Board.
While accelerator design and modelling will move in part to workstations, there will still be a requirement for use of the central computer services. Substantial use of computers of the power of the Cray will be required for design of complex accelerators.
Staffing Requirements (not including Fellows etc.) | |||
Area | Present | Future | New Posts |
Controls | 134 | 134 | 01 |
Databases | 15 | 15 | 02 |
General Infrastructure | 10 | 15 | 53 |
Note 1. | The need for increased effort in end user applications |
will be compensated by rationalization elsewhere. | |
Note 2. | Reallocation within the Accelerator Divisions. |
Note 3. | Possibly obtained by rationalization across other areas. |
The numbers in [Ref.] assume that there will be strong central support from DD. This support is currently used in the fields of: networking, ORACLE system support, CSE including PRIAM for microprocessors, MIS including PC--Shop and applications development, and central computing support on VM.
It is important that the accelerator divisions can count on the continuity of this central support and indeed on its consolidation to take these needs more into account than in the past. If not, the numbers in the table would have to be increased and an inefficient situation would result. The division between central support and local user support is always hard to delineate and must be decided according to the individual circumstances of each application. Nevertheless, the benefits and savings are such as to make the effort worthwhile.
Strong central support for the computing needs of the accelerator divisions must be consolidated. The level of support and its interaction with local support must be decided on a case by case basis. |
No additional money above current levels is requested in the working group report.
Computing for Engineers is a relatively young subject at CERN. Whilst the authors of the 1983 Green Book were aware of the growing problem of providing for CERN's engineers, there was no separate working group assigned to this area, and the associated topics were but lightly considered. Five years later, the extent to which the engineering community at CERN has come to use computers as tools, and the vastly increased sophistication of its needs, has led to a specific working group on this topic and to a separate chapter in this report. It has also led to strong pressure from the engineering community for an increase in the organized computing support which they receive, and which today is at a rather low level of manpower.
In this time, an Advisory Committee on Computing Support for Engineering, (ACCSE), prepared a report for the CERN management. [Footnote: P--G. Innocenti, "Report of the Advisory Committee on Computing Support for Engineering", 30th. September 1987. ] Following discussion of this report, CERN has decided (June 1988) that a plan for coordinating and enhancing the facilities for engineers be implemented. The resources for that implementation, and especially the manpower requested, remain in large part still to be found, and hence the implementation is still incomplete.
The working group report is in some sense a later review of the material covered by ACCSE. The requirements and requests upon the laboratory in the two reports are thus similar and the CERN management should find no surprises. The working group divided itself into seven areas and sub--groups as follows:--
O Support for Microprocessor Users O Structural Analysis O Software Support for Field Calculations O Analog Electronics CAE/CAD O Digital Electronics CAE/CAD O Database Support for Engineering O Computer Aided Mechanical Engineering and Related Fields
The working group report consists largely of the seven separate sub--group reports. Nonetheless, there are two related main themes which emerge from the report as a whole, and form the basis for recommendations:--
Experiments and accelerators depend on front--line technology that can only be achieved through the use of CAE tools. CERN is late investing in this area and is not yet investing enough in proportion with what is spent on the more traditional areas of computing. |
A large percentage of CERN staff were educated before computer--aided tools became part of the syllabus. If CERN is to gain full benefit from CAE techniques, staff at supervisory as well as technical levels will need training in modern design methodologies. |
Microprocessor support was set in place before some of the other engineering support activities, and as a result there is a clear message that the current activity is a good model and should continue in the same spirit in the future. There are a number of technical recommendations, and a request for more resources at the level of 2.5 more people in central support and an upgrade of the central support computer costing 500 to 700 kSF.
Historically, a number of structural analysis packages have been installed on the central computers, but the present trend is to use either CASTEM [Footnote: from CEA Saclay.] or ANSYS [Footnote: from Swanson Analysis Systems Inc.] in particular. Support for the thirty or so users is on a somewhat ad hoc or goodwill basis. Besides recommendations of improved bridges and integration with other areas such as field calculations and CAD systems, the major request here is for user support at the level of one post (programmer/engineer) in central DD support.
A story similar to that above is to be found for field computations. A number of packages have been installed over the years, but the major problems with them, (lack of pre-- and post--processing, and lack of good 3--D programs), are being solved by the introduction of TOSCA [Footnote: from the Vector Field (VF) company, Oxford.] and ANSYS. These commercial packages tend to be expensive by academic standards, and an annual budget of the order of 100 kSF is required. A larger problem comes, as above, from the request that the present 50 users need central informatics support at the level of two persons.
This area of support has been particularly neglected and merits urgent attention.
The major message from the working group is that it is high time that the laboratory started working seriously towards microelectronics, and got itself better organized in this area. Training has a capital rôle to play in introducing this new technology. Also required are centrally supported modern CAE packages with good handling of exotic components. The establishment of relevant model libraries for these tools is very important.
The subject is covered in a great deal of detail in the sub--group report and not repeated here. The messages too are similar. The conclusion is perhaps that this is mainly an organization question, and one in which one needs continuous investment in money and manpower.
The usefulness of database management systems (DBMS) has been generally accepted by engineers. The rapidly expanding user community and number of applications around the ORACLE relational DBMS system suggests that ORACLE will continue for a long time at CERN, there being no significantly better product on the market at present.
There is a strong request coming mainly from the accelerator divisions to provide a database service 24 hours a day, seven days a week. The intimate scheduling of the central computers to fit with the accelerator division needs (as opposed to the needs of the experimental physicists) does indeed have new implications on the running of the central service, and these need further discussion and study.
The major message from this area is that an application support team is needed to complement the existing DD--based database service support. Its tasks would include coordination of application development, database service management, application development expertise, data organization and other tasks. The bulk of the application development effort itself can continue to be provided by the engineering groups concerned under the guidance of this team.
Computer Aided Design (CAD) made a major step with the decision in 1982 to install the true 3--D solid modelling system EUCLID from Matra Datavision, and this powerful design system has been supplemented recently by a commercially widespread and simpler drawing package AUTOCAD, running on IBM--PC compatibles. This aspect of design is thus reasonably well covered technically but lacks support manpower (especially organized support for AUTOCAD).
The strongest new message coming from this area is a need for better integration of engineering tasks, linking drawing tools to databases and also to structural analysis.
Note 1. | In the non--DD column, only those numbers are shown which affect the |
recommendations. Indications of the other numbers may be found in the | |
ACCSE report. | |
Note 2. | The range of numbers recommended for Digital El. CAE corresponds at the |
low end to maintaining the status quo and at the high end to satisfying the | |
estimated future needs of the lab. | |
Note 3. | The doubt in the non--DD effort for Digital El. CAE corresponds to the |
lingering uncertainty concerning P--CAD support. This should be resolved. |
Clearly, the provision of the increased staffing is an on--going exercise which was recently reviewed by the laboratory management.
The increases will come partly from internal reassignment. A small number of internal posts have already been allocated and are, in part, being filled. There is a transition between the situation where engineers write their own code and one in which they become users of more sophisticated commercial packages for which centrally--organized support is necessary. It remains to be seen whether reassignment will be successful in providing all the necessary expertise in these areas.
The anomaly of having a service (Microprocessor Support) funded entirely with
investment money should be noted. This situation should not be allowed to occur again in
the future.
Management Information Services clearly fall into the category of the computing infrastructure of the laboratory, having a greater responsibility towards the correct administrative functioning of CERN itself than towards the provision of services for the physics community. Yet these two rôles cannot be entirely separated, since the physicists, engineers and those working closely with them need a computer service for their administrative and management work albeit in some cases of a specialized nature. In practice, certain areas of the MIS unit at CERN are also the only, or at least major, providers of certain important services to physicists and engineers, e.g. processing of large scientific documents, PC support and the PC shop.
Historically, MIS at CERN has been attributed a relatively low priority and central investment. This, together with a lack of a key overall strategy, has led to the existence of a wide spectrum of independent and heterogeneous solutions both at the level of the word processing equipment distributed around the site, and also at the level of the major applications running on the administrative data processing computer (ADP). The continual need to create interfaces which solve specific problems between subsets of inhomogenous solutions, and the frequent discovery of multiple uncoordinated solutions to the same CERN problem, are good examples of the inefficiency that has been created by this situation.
This being recognized, a senior management committee, SCAIP, was created in late '85 and decided to initiate an MIS Task Force in '86. This led to the creation in '87 of the MIS unit in DD Division. SCAIP continues to oversee MIS policy, and an MIS Board, consisting of representatives from all the divisions, provides technical coordination. The latter formed a natural basis for the working group producing input to this report. This group was joined by two participants from DESY.
The working group's report reflects the established programme of work of the MIS unit, and the resource levels required to carry it out. These have been discussed several times by the CERN management during the unit's short evolution, and this particular report has few surprises. The overall theme is one of continuity of the established programme and structures. This being said, it must be emphasized that not all of the problems have simple solutions and that the unit has not yet reached the staffing level required to carry out its programme of work. In addition, pressure from the CERN Member States for improved efficiency of the CERN administrative procedures and clearer presentation of the CERN accounts has increased the demands on the initial programme, and emphasized the need for an upgrade of the already saturated ADP computer.
In broad terms the MIS activities at CERN divide into two areas:--
The ADP computer is small, an IBM 4361, installed 5 years ago and completely saturated. Response time for transactions and enquiries during the working day is unacceptable even though batch applications run outside prime shift. For historical reasons applications run under two different operating systems. There are three data base systems, none of which is Oracle, and three transaction systems. Despite serious attempts to homogenize the user interface, one is unable to hide the "context switch" that occurs when passing from one application to another. The financial data base system and the purchasing/receiving system are different and use separate data bases. Interfacing links have been built by CERN, but cannot conceal the differences.
The second area of office systems covers several areas. The scientists have long used the central machines for their office tasks such as mail, scientific text processing and document storage. Despite the evolution of workstation software in this area, many of these tasks will remain in part on the mainframe for some time. The two latter tasks are the responsibility of the current MIS unit.
The administrative office systems began as stand--alone word processors, independently installed by independent entities, and have formed clear barriers to document interchange and a laboratory--wide system. Scientific symbols and mathematical text processing have been needed not just by the scientists but also by the secretariats. Scientific document exchange with other laboratories and with the publishers of scientific journals is another requirement. Scientists visiting CERN want to find the text system they are used to at home. Whilst stand--alone word processors are no longer purchased, document markup standards begin to ease the situation, and Macs/PCs are provide acceptable powerful solutions, there is still no unique solution that provides all the requirements. The present mixed and inefficient situation has only just started to be resolved.
The investments in this area cannot be ignored. It is estimated that over 7 MSF was spent on word processors, which today still cost a recurrent 0.7 MSF in maintenance. The Mac/PC base is estimated at 20 MSF of capital investment expanding at 5 MSF per year.
Current and future office applications focus on four unavoidable operating environments into which all present effort is concentrated, which are IBM with VM/CMS, DEC VAX with VMS, PC--compatibles and MS--DOS, and Macintosh with MacOS. These are linked by the CERN Ethernet communications infrastructure. Obsolete systems will be phased out gradually.
Note that each of the above has its own particular rôle with the CERN computing picture, but that the overwhelming choice for a desktop system by those not constrained by the needs of other environments is the Macintosh. Unix systems may bring future possibilities for simplification, but it is possibly in this area above all that the manufacturers will see the need and the chance to put "added value" into their Unix products, especially in the area of the human interface.
The working group report analyzes the requirements in section two of its report and
produces a list of required services to meet the objectives it has set, namely to:--
"
" They list 17 recommendations, separated into three main categories which may be found in detail in their report, and which lead here two summary recommendations:--
The ADP Computer should be upgraded, and its operation integrated with the CERN Computer Centre. |
The MIS environment at CERN should be modernized and made coherent. It should include a CERN corporate data model using a unique DBMS (ORACLE). This implies the replacement of the currently used major corporate applications. A comprehensive electronic forms handling system should be implemented and EDI should be used, where possible, for corporate applications. |
The total resources needed to implement this plan as far as they are known and can be estimated are summarized below. The numbers are approximate, and in some cases depend on the outcome of a study. Some of the money is financed by user divisions rather than MIS directly (see the working group report for details).
The theoreticians at CERN have traditionally used computers somewhat less that the other research divisions, although this situation is now changing. A review of their present computing requirements reveals two main themes, namely:--
In order to carry out their work in a satisfactory manner nowadays, the theoreticians make use of electronic mail, scientific text processing and other general tools. The CERN Theory Division is, however, somewhat humbly equipped even in the number of terminals available. It is generally agreed that the equipping of all offices of the Theory Division with terminals, PC's, Macintoshes or workstations, together with ready access to printing facilities is of high priority. This can be achieved to a large extent by the provision of modest financial resources.
Certain software packages have found particular acceptance within the theoretical
world, notably the
Recently developed techniques for solving basic non--perturbative problems in QCD by use of Monte Carlo simulation on the lattice require powerful computing facilities such as vector supercomputers or parallel processors. They require such large amounts of computer time that they could keep any number of powerful computers busy all of the time.
It is recalled that the computing resources in the Centre are there primarily to assist the experimental programme. Furthermore, in the general spirit of the one--third : two--thirds rule, this type of computation can just as well be done outside CERN.
On the other hand, one should distinguish between the development, debugging and testing phase of this research and the major simulation runs that consume the computer time. CERN theorists are involved in large international computer simulation collaborations. It is true that these collaborations have succeeded in finding, using their own initiative, the computer time they need at other installations. However, if CERN does not provide a certain level of facilities to match international standards, (at least for the development phase), its theorists will be unable to participate fully in this work and to attract others concerned.
There is therefore a question of policy to be resolved here, as to what fraction of CERN's central resources are to be devoted to this kind of theoretical work.
It is recommended that such theoretical work should have access to CERN's central computing resources for development and debugging but that the fraction of resources allocated should remain modest, of the order of 10% of the total available. |
The financial resources required to equip the Theory Division with normal computing infrastructure are modest, of the order of 300 kSF. Nonetheless, this can probably not be handled within the normal operating budget of the division.
The requirement to find manpower at the level of one person in a user support rôle essentially for Theory Division is a request that comes on top of existing other demands on an over--stretched area. Note that this form of user support spans the boundary between MIS and the computer centre user support group.
A number of general points come out of the above review which are not specifically brought out in any section. Two particular examples are staffing and training.
The laboratory is seriously under--staffed for the computing demands it is already asked to fulfil. The working groups make large additional demands for staff. (On face value, they ask for around 100 posts). Furthermore, in view of the overall compression of the CERN staff complement that is in progress, it will be particularly difficult to make the necessary increase in informatics staff numbers in the short term. The committee therefore feels that:--
It is essential to agree on a staged plan for an acceptable increase in informatics staffing levels across the whole laboratory over, say, five years. |
A small, but valuable, contribution to alleviating the staffing problem is being made through coopérants, joint projects with industry, etc.. Some increased activity of this nature may be possible, although it is clear that the essential part of the problem cannot be solved by this means.
We recommend investigation of further alternative methods for informatics staffing,
such as the appointment of technical associates from companies and universities, and
increased involvement of graduate students. Joint development projects and staff exchanges with industry and academia, especially in the fields of advanced electronics and computer science, should be actively encouraged. |
The success will vary with the area. Some areas can probably not be filled by re--training. Nonetheless, it is essential to establish a major programme to this end if CERN wishes to escape from this chronic staffing constraint. New attitudes and extra resources will be required to produce the desired results.
CERN must set--up an organization devoted to a massive training and re--training of the existing staff in informatics technology. |
In an associated area of training:--
Progress in the field of computing is so rapid that CERN must devote a fraction of its staff resources to work on advanced development projects that are only expected to show benefits in the medium to long--term. It is also vital that the long--term CERN staff involved with computing are kept well informed of the latest techniques. |
|
This Appendix is a rather brief resumé of the computing facilities in place at CERN at the beginning of 1989, intended to serve a base--line to the discussions contained in the main report. A much higher level of detail can be found in the DD Report CERN/DD/89/9.
The main batch and interactive services are provided by a six processor IBM 3090/600E with six VFs and a twin processor Siemens 7890 S (Fujitsu M382). These machines provide a total scalar processor capacity of some 52 CERN units, [Footnote: the CERN unit of computing power is historically the power of an IBM 370/168 or a VAX 8600. As a rough indication, this may be taken as 3 IBM MIPS or 4 DEC VUPS. ] and share access to 25 tape drives, to 32 cartridge drives, and to 205 Gbytes of disk space. The main operating system is VM/XA SP2 with HEPVM [Footnote: HEPVM is the name of the collaboration between High Energy Physics sites which seeks to make such modifications and additions to the VM operating system that are important for HEP as part of a common effort, and which thus eases the task of moving physics code between sites. The most essential additions concern the Batch Monitor written originally at SLAC. The key member sites are CERN, IN2P3, RAL, SACLAY, and SLAC, and there are around 25 collaborating sites.] additions. The MVS/WYLBUR operating system used for the previous ten years is being actively phased out, together with its key peripheral device, the IBM Mass Storage System (MSS). The IBM system has 5000 registered users of whom more than 2500 use the system in one week. At peak hours there are around 600 connected users of whom close to 300 are active each minute.
A Cray X--MP/48 (4 processor, 8 Mwords memory) with a solid state disk of 128 Mwords and 45 Gbytes of disk space runs the Unicos [Footnote: Cray's multiprocessor version of Unix] operating system. It provides around 32 CERN units of scalar processing power. Whilst this machine is clearly targeted at codes suitable for high levels of vector performance it is also running some scalar data processing codes from the experiments. The Cray is connected directly to the above IBM tape and cartridge drives. The Cray has around 200 registered users.
The exploitation of vector processing facilities on the Cray and IBM for the codes that concern experimental physics is the subject of much active work. The vector speed--up for such codes has been traditionally small, and new ideas are needed. The potential gains are nonetheless high.
Major interactive services are also provided under VMS on a VAXcluster which contains five powerful VAXes comprising 8800, 8700, 8650, and a number of smaller models. The key features of this popular service are compatibility with VAX systems in the experiments and the home institutes, and the suitability of the user--friendly operating system for development work. There are nearly 3000 registered users of whom 1500 use the system in one week. At peak times the two VAX machines dedicated to the general service have 200 connected users.
Besides this general service, the VAXcluster provides specialized services for LEP Project databases (based on Oracle), and mechanical CAE (based on Euclid). Another VAX, an 8530, runs an Ultrix (Unix) service for a number of specialist areas including microprocessor support.
The INDEX terminal switching system connects around 2000 terminals to 200 hosts. No further investment in this facility is planned, but it remains a primary route for connecting terminals. Around 30 bridged Ethernets link some 1000 computers and other devices across the site. Terminal concentrators on the Ethernets provide for new connections and the eventual phase--out of INDEX.
The home--developed CERNET still has some 50 computers from several different manufacturers attached to it, but is being closed down. The on--site X.25 network is connected to the public packet switched networks, and to some 12 external leased lines, including one satellite and one under--ocean link to the USA. The Swiss EARN node is located at CERN, and is attached to a further 12 leased lines.
Most large experiments have at least one powerful 32--bit computer, typically a large VAX, and a number of smaller machines, usually in a cluster, used for a variety of tasks including data acquisition during detector tests and calibration runs. There are at least 100 VAX and microVAX computers, and at least 20 Norsk Data computers (including six 32--bit Nord--500s) being used at experiments.
The PS control system consists of two Nord--570 computers and around 30 Nord--100s. The Linac and Lear are controlled by a VAX--based system, linked to the above. The SPS and LEP control systems are largely integrated, the main hardware consisting of two Nord 570s, around 80 Nord--100s, and about 100 machines of IBM--PC/AT architecture. The SPS network hardware is based on six interconnected stars, built under contract to CERN specifications by TITN about 10 years ago. The LEP network hardware is based on 16 IBM Token Rings. Workstations, notably Apollos, are being used increasingly for a variety of tasks.
CERN's ADP services run on an IBM 4361 Model 5 computer, running the DOS/VSE operating system under VM--SP. The machine is saturated and the interactive response is unacceptable. The one Gbyte of disk space used to hold the active copies of the main corporate databases, is insufficient, e.g. purchasing records older than 3 months are archived, at considerable inconvenience to users and ADP staff.
CERN has inherited a large number of diverse office computing systems. In terms of numbers of users the biggest system is the ND Notis, but there are also AES, Wang and Philips systems in use.
The best estimates indicate that there are at least 2000 Macintoshes, and at least 800 IBM--compatible PCs at CERN today, and they are all spread widely throughout the organization. Many of these have been purchased by outside user groups.
There are over 120 Apollo workstations, and at least 250 VAXstations installed currently at CERN. Of the Apollos, some 65 are on a main network, being used for software development and physics analysis by physicists and by some DD staff. A further 20 are used in LEP controls, and around seven are used for electronic CAD design. The VAXstations are also mainly used for software development and physics analysis.
This section describes the staffing levels and the money spent on computing across the laboratory over the past decade. The numbers used in this Appendix are taken from the CERN accounts. Where the numbers have been corrected to present day figures, the correction factors used have been the appropriate CERN official indices.
This being said, the CERN accounts do not contain an explicit record of the "costs of computing" as such. Indeed the term "costs of computing" needs definition in such a discussion. Three examples in common usage are, the costs of the computer centre, the full costs of the DD Division, or the costs coded in the CERN accounts under headings labelled "computers".
The following strict definition is used in order to obtain consistent numbers which may be compared across the years, and in order to handle computing expenses from all divisions, not just DD. It is assumed that a reasonable picture for material costs is given for operation money by the sum of the expenses coded under 251, 252 and 253 in the accounts (data handling: maintenance, general expenses and consumable items), and for computing capital purchases by the code 351 (data handling equipment: capital outlays). The precision of the numbers is thus dependent on the use of the correct code by the initiator of an expenditure. Computers or other data handling items purchased under other names and codes are clearly not recorded in these numbers. This definition does not cover money spent e.g. on buildings or power supplies for computers, on electrical services, on laboratory equipment, or any form of divisional infrastructure such as travel money. The numbers for DD Division are well understood and provide an indication of the strictness of the definition. In 1988, this method derives a DD total of 22.3 MSF, which can be compared with the total divisional expenditure of 25.9 MSF. This ratio varies over the years concerned between 84 and 92%.
From the numbers in [Ref.] one can see that the total CERN computing material costs thus defined, (adjusted to 1988 prices by using the CERN material indices), have stayed essentially constant over the past decade, as have the DD computing costs. This leads to a rather constant ratio, showing that DD represents on average 68% of the CERN total material computing costs.
The personnel costs are well defined for the services provided by the Computer Centre, and for the DD Division as a whole. However, the true staff involvement in computing across the rest of the laboratory is not recorded, and cannot be reliably determined. For example, the distinction between physicist or engineer and computer support staff can be difficult. Similarly, it is not easy to understand what fraction of the accelerator control staff should be considered as computer specialists.
In order to estimate the total personnel costs in a manner that allows a comparison across the years, a major assumption is made that the ratio of personnel costs to material costs for the central computers is also applicable to the laboratory as a whole. This ratio varies between 0.61 and 1.09 and averages 0.75. [Footnote: In 1987, a year of low DD material costs, this leads to an obviously incorrect personnel cost, demonstrating the limitation of the method. ] The personnel costs produced in the above manner have then been corrected to real 1988 prices.
shows that the sum of personnel and material costs for computing at CERN thus defined
have changed little over the past 10 years, and average around 55 MSF. This represents
some seven percent of the total CERN budget.
The strict definition used above produces certainly a lower bound on the true costs of computing in the laboratory. By definition, the method used does not reflect the growing number of staff outside the DD Division who have been increasingly involved in computing activities. An alternative approach is offered here as an upper bound.
One assumes that all activities within DD Division relate to computing. The accounts show that the full personnel costs of the division have stayed, (with necessary corrections for changing composition and in 1988 figures), remarkably constant at 25.0 MSF/annum over 10 years. Making a rough estimate, one finds that the number of staff outside DD who are principally engaged in computing activities, (programmers in experiments, accelerator control etc.), is of the order of 200 people, and hence is similar to that working within DD. The total personnel budget involved is thus of the order of 50 MSF.
The total materials costs for all activities in DD in 1988 were 25.9 MSF. Assuming that this represents of the order of 70% of the total CERN computing costs, (see above), one may estimate the latter at around 35 MSF. This leads to a number of around 85 MSF for the current total annual CERN costs of computing, broadly defined.
In order to record the base line of expenditure for the review and recommendations in the report one may make the following breakdown. The money allocated to the DD Division for projects, essentially the whole capital investment of the DD Division, has averaged 10 to 11 MSF/annum over the past few years. This money can be divided into a pattern of expenditure the major components of which are:--
In addition, the Cray computer is funded via the operational budget on a lease/rental contract at the level of 3.6 MSF/year.
In the following we review the capacity potentially available for processing of data from CERN experiments in European centres outside CERN. The term data processing is understood here as requiring a minimal set of organized resources, including not only cpu power but also directly connected reliable tape/cartridge units, and operational facilities such that the processing power may be used a major fraction of the week. In the spirit of this definition, many physics departmental computers and workstations are believed to contribute more to the physics analysis than to the data processing of an experiment.
The numbers given below in CERN units [Footnote: The CERN unit of computing power is historically the power of an IBM 370/168 or a VAX 8600. As a rough indication, this may be taken as 3 IBM MIPS or 4 DEC VUPS.] are the result of a survey of physics users and staff of the computer centres known to us. The list is an integration of the centres declared to COCOTIME, those involved in HEPVM, and centres declared by the experiments to a MEDDLE survey as a source of capacity. There are reasons to exercise care when reading these numbers:--
We have thus concentrated on major centres, set up for production, having tape/cartridge units, disks and operational facilities to run overnight. All with close connections to HEP have replied. Some with distant relations have not yet given answers. Beyond the major centres there are smaller physics department machines and physics' share of the general university machines, whose total capacity is estimated from those that are well known. (Typically one or two CERN units each).
The capacity available generally at Universities and research institutes is growing, including that which is available and useful to HEP. Thus the table below will soon be out of date and should be treated as giving lower limits.
The following lists the capacity that could be made available in the European data centres for processing of HEP data originating at CERN. It excludes work for theoreticians, data from DESY, CERN data processed outside Europe, nuclear physics data, and "private facilities" at CERN.
In the context of the report 'Computing at CERN in the 1990--s' various working groups were mandated to produce recommendations in the area of 'Computing for Experiments'. They have condensed their findings in two separate report parts, one dealing with questions of data acquisition problems (real time) the other containing all other aspects of data processing. The recommendations are based on the assumption that the CERN physics programme will be dominated by LEP experiments with 'Z--like' data rates until 1995, that the SPS collider programme runs well into the early 1990--s, that other experiments like the LEAR or the fixed--target programme will continue, and that serious preparation for a new hadronic collider programme (e.g. LHC) will also be part of the future scenario.
The recommendations of the working groups can be grouped into the following three broad categories.
The expected evolution of the needs of experiments and of the market offerings (i.e. new technologies) should result in a substantial boost of installed CPU capacity and mass storage, both at CERN and in member states. A major impact of modern workstations is predicted, resulting in serious needs for interconnection and software support for distributed systems.
The projections are high in capital expenditure and manpower. The working groups want to stress their opinion that present levels of funding and staffing are at a marginal or inadequate level. We believe that computing expenditures provide at least equally significant potential in added physics returns when compared to spending for other sectors such as accelerators or detector construction.
Experiments have dramatically changed in scale over the last decade, and data handling methods used outside our community have developed significantly, too. The basic attitude to the use of computers in experiments has not evolved in a similar way. More professionalism in software design, and the move towards commercial software will have to translate into new and more centralized support for the HEP community. This concerns coordinated product evaluation, the conduct of pilot projects, HEP--wide licensing strategies in many areas, some HEP--specific software developments in areas not covered by the market, and a marked effort of user education and product documentation.
The reports also suggest several areas in which research and development work is thought necessary (the understanding of 'research' here is in the direction of technologies and methods existing in industry or academia, but largely unknown to the HEP community). Most notably this concerns 'parallel computing' and 'real--time data handling for high event rates'.
We believe strongly that the CERN management should provide clear guidelines for this critical technical area. There does today exist unwanted duplication and sometimes frustration in the collaborations, who have to tackle general problems with inadequate resources. While initiatives must come from the physicists, general development and research activities cannot be pursued uncoordinated without introducing inefficiencies. In some areas, lack of a general strategy may jeopardize our physics programme.
Most recommendations made in the working group reports concern the needs of the community of physicists participating in the CERN physics programme. In some areas such as mainframe computing aspects, present policies provide guidelines for the relative sharing between CERN--based and outside, where 'outside' includes regional centres and individual university institutes.
Such guidelines do not exist for many other recommendations, particularly in the area of application--oriented support such as data base systems, data modelling, software engineering tools, hardware description languages, or graphics.
The CERN management sees its role mostly under the angle of the host laboratory, and has not shown major initiatives to address community problems in the area of computing. The HEP community, on the other hand, has as only representative body in computing matters the 'High--energy Physics Computing Coordination Committee' (HEPCCC).
The working groups strongly suggest that a mechanism is created that permits the community to take a decisive and responsible part in shaping the future landscape of computing for HEP in Europe. This could be derived from a body such as HEPCCC or possibly ECFA. Until such a global solution is found, CERN should take the lead where necessary.
The working group is of the opinion that CERN is entering a period of steady and predictable evolution as far as the real--time computing needs of the current generations of experiments are concerned. The major components of these experiments have been created, but will need running in. Continued support will be required to make the real--time systems of these experiments respond to the evolving demands of the physics programme.
The major challenge appearing on the longer--term horizon is that of constructing trigger and data acquisition systems for very high rate hadron colliders such as the LHC. If CERN chooses to pursue this ambitious program, a great deal of research and development work is required. This work is too complex to be left until the actual time of formation of experimental collaborations, and should be starting soon. We should clarify that R&D in this area relates to learning techniques and methods in use outside HEP, and will have to make use of all commercially available products and services.
The extreme environment of future hadron colliders requires major advances in almost all areas of detector construction as well as triggering and data acquisition. This forces a tight coupling of the issues of detector design, detector digitization, data compression and triggering. It seems imperative that CERN consider means of organizing and supporting the required long--term R&D in the widest possible context, integrating most or all existing expertise in the HEP community. |
Some of the general features of an LHC experiment, and the implications for data acquisition and triggering, are listed here:
There is a clear need for a better understanding of this environment, and of the methods to tackle the associated problems. Topical workshops as they have been or are being organized provide a useful starting point, but the next steps involve working with sophisticated detector and physics models, and should then lead into a series of coordinated R&D and pilot projects.
Underlying all of this will be the need to design, construct and maintain a system of the required scale. The solution will require familiarity with methods applied in other areas of complex system design, as in industrial digital systems. The introduction of such new design methodologies will take a considerable amount of time and will not be inexpensive; the necessary high--performance workstations and the corresponding software packages are both costly. There will certainly be a long and painful learning period.
Serious investments will be needed in the training of engineers, programmers, and physicists, in order to take advantage of the computer aided design methods which must be used to construct such a system. |
These methods are very powerful, but there will not be any instant results. The creation of pilot projects has proven to be an effective means for structuring such R&D. By pilot projects we understand projects which have some use beyond being strictly pedagogical, but which do not require the immediate mastery of advanced techniques, thereby leaving room to explore the new subject area.
Beyond the design methodologies mentioned above, pilot projects would also seem particularly important to acquire familiarity with VLSI technology, which we believe will be used extensively for both analogue and digital electronics of future experiments.
We recommend to set up coordinated pilot projects for introducing new high--level design methodologies in both software and hardware, and for acquiring familiarity with the application of VLSI techniques, both analogue and digital, to future experiments' real--time problems. |
To achieve this goal, a better understanding is needed of the commercial market, of other applications, and of our own medium--term evolution. This would comprise familiarity with systems like Unix (trademark of AT&T) and associated real--time kernels, and future high--bandwidth busses, large systems with distributed functions, data base support, user interfaces, fault tolerance, etc.
We believe that improved mechanisms are required to evaluate existing commercial products and the real cost of in--house developments. We also think it extremely important and even urgent to find ways of taking community--wide decisions about the acceptance and general support of such 'standards', be they commercial products or complementary developments inside our community. |
An estimation of future computing needs for the CERN community requires predictions of the future of the CERN physics programme. It has been assumed that LEP runs mainly on the Z until the end of 1991, after which running at higher energies is interspersed with additional running on the Z for several more years. The SPS Collider and fixed--target programmes are expected to remain active well into the 1990s, and LEAR experiments have firm plans to take data until at least 1992. Finally and more generally, it is assumed that European physicists will pursue a vigorous, but as yet unknown, programme of physics at CERN beyond the currently planned limits of existing experiments. By 1991--2 the physics programme of CERN will require within Europe at least 500--600 units of CPU power, the management of data amounting to hundreds of thousands of magnetic tapes, and well over 1000 Gigabytes of direct access storage. The number of workstations, each with about one unit of CPU, will approach the number of physicists in HEP. To ensure full university participation in this physics programme, users at home institutes will need access to both CERN and their regional centres at the speeds possible using modern fibre--optic technology. The European HEP community has major experimental programmes at DESY and at the Gran Sasso Laboratory, in addition to those at CERN. The quantitative conclusions of this report cover only the requirements of the CERN--based programme. The qualitative conclusions, on networking and distributed computing, are almost certainly applicable to the other programmes.
This report makes a number of recommendations which should be addressed to the management of European HEP rather than to CERN alone, or to laboratories, computing centres and universities independently. No existing body combines the managerial and detailed technical competence needed to implement the recommendations, although HEPCCC and/or Restricted ECFA may be in a position to initiate the creation of the necessary structures.
European HEP requires a body which can provide continuing managerial and technical co--ordination of all aspects of the computing environment. |
Workstations are single--user computers with high resolution graphics screens, and capable of running and debugging the largest programs used in HEP. In the period up to 1995, the number of workstations used by European HEP will become comparable with the number of active physicists. Full exploitation of workstations will require that each workstation can access data located at CERN or at regional centres at 40 to 400 kilobytes/second per CERN unit of computing power in the workstation. This requirement might well require a carefully organized hierarchy of servers.
A coordinated effort is required on the part of CERN and regional centres to propose, implement and continuously develop, a strategy for distributed data management for European HEP. This work should be supervised by a technical committee. |
Management of workstation system software and communications will become a task comparable to the management of the mainframe computing service at CERN and at regional computer centres. As far as possible this management should be provided by the computer centres for the benefit of CERN--site and university users. Workstations applied to real--time tasks should be included in these considerations, wherever possible.
A service should be set up at CERN mandated to coordinate the management of systems software and communications for a defined but evolving range of workstations on the CERN site. Serious consideration should be given to offering a complete support service, rather than the coordination of manpower supplied by collaborations and other users. |
Similar services supporting local and university users will be appropriate for many regional centres. A committee should coordinate the activities at CERN and regional centres. Differing data representations reduce the transparency of any distributed computing environment. Conformance to the increasingly widely accepted IEEE data representation should be an important factor in the choice of hardware.
Mass storage capabilities will be at least as important as CPU power in supporting flexible analysis of LEP physics, and will limit the scale of off--line analysis for hadron colliders.
By end 1991, CERN should have a total of at least 1000 Gigabytes of disk space and an automatic handling system for at least 40,000 tape cartridges. By end 1991, any regional centre planning to contribute effectively to the analysis of major experiments should have at least 100 Gigabytes of disk space available to experiments together with automatic handling for at least 5,000 tape cartridges. |
The cost of these mass storage facilities will be tens of millions of Swiss Francs, and even after this expenditure, resources will require very careful management.
An automatically managed storage hierarchy, from disk--files to data--tapes, is vital for CERN and will be important for regional centres. The financial value, in terms of more effective disk space use, is many millions of Swiss Francs. If such storage management remains unavailable commercially, it must be written by, or under contract from, CERN. |
IBM 3480 tape cartridges will be the main mass storage medium for CERN physics in the next five years. However, even with the improvements in cartridge capacity now expected, there will still be hundreds of thousands of active tapes at CERN.
New mass storage devices with much higher capacity should be studied as alternatives in the longer term. |
Conventional mainframe computers will continue to be vital components of the HEP computing environment. However, in the period up to 1995, there will be a change in the way they are used. By 1995, most physicists will use workstations rather than terminals connected to mainframes. Physics analysis programs on the workstations will interact frequently with the mainframes, exploiting their unique combination of massive real--time data handling capacity and substantial CPU power. Workstations will certainly allow more productive and reliable physics analysis, but are unlikely to reduce the steady growth in the demand for mainframe services. The CERN physics programme will require at least 500--600 units of CPU power by 1991--2. The physics would benefit considerably if much more power were available. Approximately one quarter of the 500--600 units should be in the form of conventional mainframe systems on the CERN site. One of the main functions of these systems will be to act as data servers and communications servers for off--site and on--site users. At the end of 1988 the mainframe scalar capacity at CERN will be about 80 units (CRAY, DEC and IBM). It is hoped that the vector capabilities of the CRAY and IBM machines will make additional power available for some Monte Carlo simulations, but vectorization is unlikely to have a major impact on other areas of physics analysis.
By 1991--2 CERN's mainframe scalar capacity should be increased to at least 150 units. |
Major regional centres should also be upgraded so that their aggregate power available to CERN--based experiments exceeds that available on the CERN site. It is assumed that parallel computing systems will be able to provide the balance of the 500--600 units at a total cost below that of conventional mainframes. If parallel computing proves to be much cheaper than using conventional mainframes, it would be beneficial to provide a total capacity substantially above 500--600 units. Conversely, if parallel computing is not successful, the capacity will have to be provided by other, conventional means.
A 'parallel computing project' should be set up as a coordinated effort involving CERN, major regional centres, interested university groups and industry. The project goals should be to do research in parallel computing relevant to HEP needs, and to offer, almost from the start, a parallel computing service to users. A system of commercially available processors adequately integrated with central data handling services should be given serious but not exclusive consideration. Physical implementations should be planned at CERN, and at some universities and regional centres. |
Parallel computing systems are also important for on--line event selection. The off--line parallel computing project should be closely linked to the development of comparable on--line systems.
The key to effective university involvement in today's collider experiments with high data volumes is fast wide area networking. Traffic between universities and 'data servers' at CERN or in regional centres is best carried by a dedicated network. The network should be distinct from a general purpose 'European Research Network', although its traffic might be multiplexed over the same low--level infrastructure. A glass fibre has a capacity today of about 34 to 140 Megabits/second using widely available drivers. Off--the--shelf technology is expected to reach 565 Megabits/second and beyond in the early 1990s.
The short term target of 2 Megabits/second between CERN and each regional centre by 1991 should be seen as a step towards installation of at least one glass--fibre--equivalent as soon as possible. University involvement is fundamental to CERN's existence; a resolution of the funding and tariff problems associated with high speed international networking should be vigorously pursued at the highest level of management. |
Countries where high speed lines are not yet available should be encouraged to speed the necessary installations to enable their full participation in the CERN physics programme. Progress in local area network technology and management is essential for the effective use of workstations. Full exploitation of wide area links above 2 Megabits/second is beyond the capabilities of current commercial technology.
CERN and regional centres should continue, and where necessary expand, their efforts to work together with industry at the forefront of network technology. |
Standards should be supported and adopted wherever possible, while recognizing that a blanket 'wait for the standards' policy is quite unacceptable.
Where necessary, data distribution by wide area network should be supplemented by, and integrated with, the physical transport of magnetic tapes.
CERN should offer a tape copying and distribution service such that tapes can arrive at regional centres and universities within days of submission of an appropriately authorized electronic request. Regional centres should also have tape copying and distribution facilities. |
Unix is already used on the majority of workstations, and is available for VAX and CRAY machines. It will soon be possible to install a version of Unix on IBM mainframes, offering, for the first time, the possibility to perform the full range of HEP computing tasks under a single operating system. The recent 'Open Systems Foundation' initiative increases the probability that these versions of Unix will appear very similar to the user. The effectiveness of using this single operating system can only be measured by making it available to groups of users, or complete experiments, prepared to use it for all their computing.
A Unix service should be started on the central IBM system at CERN as soon as adequate software is available from IBM. An initial small--scale service should receive further resources according to user demand. |
In spite of the many deficiencies of Fortran, there is no immediately obvious superior alternative. The CERN community should periodically review the developments in languages, both to add weight to its input to Fortran standardization committees, and to identify any serious alternative to Fortran. The production of reliable analysis programs for complex detectors is increasingly slow and manpower intensive. The full exploitation of complex detectors at future colliders will require the use of techniques (or languages) to generate low--level code automatically from a higher level of abstraction. No existing language fulfils this need.
HEP must organize its support for HEP--specific software more effectively.
Formal procedures are needed for determining and responding to user needs. |
Some freedom should exist to pursue good ideas in the absence of clearly formulated demand, subject to periodic user--dominated review.
Physicists would benefit from overall coordination of the software support provided by CERN, DESY and regional centres. This coordination should be effected by the creation of a Software Support Committee. |
A first task for this committee would be to assess the adequacy of current manpower levels. There is no doubt that widely available and successful commercial software, meeting HEP needs, should be bought and not re--written within the HEP community. Experience with the purchase of less mature commercial software has not been uniformly encouraging. An additional task for the Software Support Committee should be to advise HEP management case--by--case on the relative benefits of buying commercial software compared with writing HEP--specific products.
The use of data base techniques and similar access methods for managing large volumes of structured data has turned out to be a key factor in today's experiments. The basic support of Oracle, a commercial data base management system (originally called for by an ad hoc subcommitte of ECFA) has convinced experiments of the value of such general products, although this system is expensive.
However, commercial products do not normally come with a user interface of the desired level, or with the full functionality physicists request. Applications of data base techniques in experiments cover a wide range of areas: administrative directories of collaborators, calibration data, detector descriptions, data models, and many others. The requirements of experiments, nevertheless, are quite similar or even identical. Users of the past have noticed systematic inadequacies of Oracle in different domains, particularly in access speed and in ease of learning. Substantial effort was invested by experiments to overcome these problems, without resulting in generally adequate solutions.
;It is recommended to coordinate across the HEP community the definition, writing, installation, and support of data base--oriented software, for all applications common to large experiments. This concerns commercial products, physics--specific application layers to commercial products, and physics--specific products not available from outside our community. |
For some applications, such work has to maintain close links with existing or proposed solutions found for data bases in a different context, e.g. administrative data bases in laboratories, or data models used as part of software engineering techniques.
Some experiments of the generation now approaching the data taking phase have introduced software engineering methodologies and tools in the design of their experiments, mostly in the spirit of pilot projects. The overall experience with such methods is judged positive. Some aspects of software engineering, like data modelling and structured design and analysis tools are today considered vital in designing large experiments. This is particularly true for the frequent case that the collaborating individuals can not work in geographical proximity; 'vital' in this case has to be understood in its literal sense as: absence of adequate software planning may jeopardize the experiment.
;It is recommended to coordinate, at the level of the HEP community, the selection, creation, and support of tools and methodologies in the area of software engineering. This includes the evaluation of commercial products, the conduct of pilot projects, the writing and/or installation of tools, and the ensuing support tasks. |
Graphical presentations of data have been recognized by all present experiments to be a critical element in modelling and understanding detectors, data reduction methods, and physics analysis. Despite large investments, the access to graphics remains limited by the available manpower: partly this may be ascribed to the notorious mismatch between visual and algorithmical concepts; more importantly, the rapid evolution of the market in both hardware and software makes the emergence of standards in this area difficult and slow. In addition, the structure of HEP collaborations with their independent decision mechanisms governed by quite different national constraints makes even the possible standardization difficult to impose.
The decision by HEPCCC to impose GKS as general portability standard is generally seen as a success, although the specific product supported by CERN is sometimes criticized. The community's need for CERN's support in this area is considered essential, new activities are suggested to become part of this support, and more user participation is desired. The present staffing level must be called clearly insufficient.
Looking beyond the next few years, the success of GKS is possibly a partial and temporary one. The GKS interface is seen as too low--level, and standards can be expected to evolve towards more structured data (as in the PHIGS standard), better metafiles, better interactivity, introduction of surface rendering, etc. Experiments need HEP--specific high--level user interfaces not provided by GKS, and of longer lifetime than expected for GKS.
It is recommended to provide sufficient manpower with good technical expertise at CERN to support GKS, the chosen graphics standard. It is further recommended to coordinate much better the existing efforts in the graphics area across the HEP community, in view of harmonizing the non--standard graphics extensions, and with the aim of preparing today the phase of transition towards standards beyond GKS, e.g. by supporting an intermediate--level general interface for application programs. |
Many common problems of software are solved independently today by different mostly small groups. This favors creative and novel solutions, but often is also seen as an undesirable duplication of efforts. The origin is not only inadequate coordination, but also a clear lack of communication. Existing solutions, even if implemented at a high level of professionalism, rely for their spreading across experiments on oral transmission and person--to--person tutoring more than on professionally written documentation and expertly organized training courses accessible to all.
;We recommend that the problem of instructing the community in the use of generally accepted methods or tools be given very serious attention. We suggest as possible and adequate steps the hiring or redirection of qualified staff for technical writing (documentation) and for the organization of repeated technical training courses. |
Training courses should be given by the best experts, from inside or outside the community. They should have access to training techniques novel for our community but well known in professional training elsewhere, like video presentations and computer--assisted instruction.
The following points are relevant to the cost and manpower tables.
These approximate estimates are for CERN only. The resources for the whole CERN community will be at least twice those for CERN alone. Wide--area line costs are included for information; it is not implied that they should be paid by CERN.
These approximate estimates represent resources located at CERN and serving the whole of the CERN--connected community.
HEP computing needs cannot be expressed in absolute terms. The quality and quantity of the final physics output of an experimental programme changes rather smoothly as the total funding for computing, or its division between real--time, CPU, storage, networking, or application programs is varied. Ideally, this section should establish the way in which physics output would change as funding for each component was varied, leaving the management of European HEP to make appropriate informed decisions about resource allocation. It is not possible to quantify physics output without getting into fruitless controversy. Nevertheless, it is the opinion of the working group that the computing needs described above lie well below the point where further resources would bring negligible physics gains. If the resources were either halved or doubled there would be a significant (10 to 30%?) change in the potential for uncovering fundamental physics. To optimize funding for HEP computing it is necessary to compare these tradeoffs with other ways of improving physics output, such as running LEP for an extra year or even installing spin rotators. The cost of all the additional hardware (on and off the CERN site) needed to meet the stated computing needs of LEP experiments is of the order 10% of the hardware cost of LEP and the LEP detectors.
The combination of CPU power, storage and networking described above form a balanced set of resources which have, in the longer term, equal priorities. If, after consideration of the likely impact on physics output, management decides that the resource allocation should be below (or above) that needed to meet the requirements described in this report, the reductions (or increases) in resources should, in the longer term, be spread evenly. In the shorter term, two factors may lead to a need for uneven funding. The first factor is relatively trivial: hardware, particularly when bought at attractive discount levels, is usually offered in a highly quantized form. This working group cannot pre--judge whether pursuing a particularly attractive offer for a mainframe might warrant a delay in disk installations. The second factor is related to the differing development times for the various facilities. Assuming that floor space is available, the acquisition of disks, or the enhancement of existing mainframes, can be delayed until just before the time when they are needed. Other facilities, which require new technological developments or imply new ways of doing physics analysis, cannot be delayed until the last moment. The following developments must be started very soon if they are to benefit the current CERN physics programme:
2 Megabit/second Links It is technically possible to offer immediately a restricted
range of services which use the full bandwidth of such links. To develop a full range of
services, and to integrate these into the working methods of physicists in time for the
most LEP physics analysis, will require that at least some 2 Megabit/second links are
available very soon.
In addition to investments in mainframe--related services, several recommendations above advocate a more coordinated approach to problem areas common to many experiments. Many of these problems are new, due to the evolution of the market in software and hardware. Some arise because experiments have reached a level of complexity which cannot be handled by simple extrapolation from the past. The solutions of such problems today is largely left to experiments. Indeed, the suggestions to coordinate come predominantly from the physicists in the experiments trying to cope with general problems using inadequate support in staff and, sometimes, money.
The working group can not assign individual priorities to the proposed activities, be they relevant in a shorter term like the application--oriented software for graphics and data bases, or of longer--term impact like familiarization through an R&D program with digital VLSI technology for real--time data control and triggering, the introduction of methodologies in hardware and software design, or the European coordination of many of these activities. We believe that each recommendation, if followed, will make a visible contribution to the physics programme. Some of them may be mandatory for CERN's future acivities: The dependence of a possible LHC program on mastering the real--time problems need not be underlined further.
The individuals who contributed to this working group at various stages are listed below :
K.Einsweiler | (CERN/EP,Convener) | A.Bogaerts | (CERN/DD) |
R.Dobinson | (CERN/EP) | J.Dorenbosch | (NIKHEF) |
J.Harvey | (RAL) | L.Levinson | (Weizmann) |
A.Marchioro | (CERN/EF) | H.Muller | (CERN/EP) |
D.Notz | (DESY) | M.Sendall | (CERN/DD) |
H.Von der Schmitt | (Heidelberg) | - | - |
This report discusses the perceived needs of the experimental physics community at CERN in the broad area of Real--Time Computing in the period up to 1995. This subject extends from the realm of front--end detector electronics where Digital Signal Processing using commercial processors is playing an increasing role in the data formatting and trigger decisions, up to the level of the cluster of Online Computers (at present, usually VAXes) which provide the user interface and other higher level functions in a modern experiment. On the lowest level, there is considerable overlap with electronics topics (special purpose processors, custom IC's, etc.), whereas on the higher levels, there are many concerns in common with offline subjects (software engineering, human interfaces, databases, etc.). In these overlapping areas, we try to emphasize those aspects of the problems which are of particular importance in Real--Time Computing.
The overall structure of the report divides naturally into two sections, one discussing the likely requirements of the expected physics programme, and one discussing the implications for the future evolution of data acquisition systems. It includes some speculations about what will be necessary to support physics in the 'post--LEP' era, which we assume to be dominated by a high--intensity hadron collider. For short, we will call it LHC. We have summarized our suggestions for the future in part 2 (chapter 1) above.
Forecasting the direction that Real--Time computing will take, needs multiple contacts, pilot projects, and a decision making process which includes the experiments. There was only limited time available for the preparation of this document. This report, therefore, contains requests for further study in several areas rather than suggesting direct actions.
The first requirement for this discussion is a Reference Model, containing our expectations for what experimental activity will be taking place at CERN in the period of 1989--1994.
The following is a summary of the major areas of experimental activity.
We know in principle how to solve most of the problems of these experiments. We can forecast some trends and extrapolate from present systems. It must be emphasized however, that a large effort is still needed on these "conventional" data acquisition systems. This must cover commissioning, running in and improving the data acquisition systems of the LEP experiments, and of those LEAR, fixed--target and collider experiments planning new or upgraded systems. The systems will need to track evolving technology and user requirements. In addition there will be new experiments and tests proposed, amongst them the activities in preparation of LHC, as recommended below.
Real--Time computing is driven by the needs of actual experiments. This makes it especially difficult to predict the long--term needs. The discussion here focuses on a future scenario involving a high luminosity hadron collider (LHC). It should be noted that this is a demanding experimental environment, with extreme needs, particularly in the areas of front--end electronics and trigger processing. The nature of the subsequent discussion thus depends very strongly on the detailed design of the experiments. The overview presented here will be very schematic, attempting to highlight some of the problem areas.
We start with a brief description of the characteristics of the machine and detectors,
essentially summarizing the recommendations of the La Thuile study on Trigger and Data
Acquisition (contained in the CERN Yellow Report 87--07). We restrict our discussion to
the
The standard machine parameters involve a crossing interval of 25 nsec at the design
luminosity of
It is clear that these LHC requirements exceed the current state of the art by several orders of magnitude.
The performance requirements for the data acquisition system can be similarly summarized (again, the discussion is for a calorimetric detector - a more complex detector will surely introduce additional problems) :
A further comment should be made about the simple monolithic description of triggering and dataflow presented above. In an experiment operating in the complex physics environment of LHC, there will be many different physics analyses proceeding in parallel. Many of these will want to examine processes which occur at rates much higher than the output rate of several Hz normally discussed for permanent storage. It will be quite feasible to perform some of the DST generation in the LEVEL3 farm, vastly reducing the amount of information to be recorded for some classes of events. One would expect to see a migration of analysis from the offline environment (after permanent storage) to the online environment (before permanent storage), as experience accumulates, in order to cope with the data flow problems inherent in analyzing high cross--section physics processes. Thus, the LEVEL3 farm becomes a hybrid computing facility, containing elements of offline and online environments, which will need to support physics analysis as well as event filtering. This broad overlap with what is traditionally called offline computing should be examined in more detail. Since the bandwidth for transferring data to processor memories is much higher than that for permanent storage, this may be one of the most important tools in performing analyses on large event samples, avoiding the bottlenecks associated with the full offline computing environment.
This group has not carried out a detailed analysis of the reference model, but it is clear that there will be a great deal of effort expended on making the major new systems work (especially those of the LEP experiments), and learning from their deficiencies. Much of this effort should be invested in such a way as to be a basis for future LHC systems, which in addition pose new and difficult problems. In discussing LHC, the natural bias of physicists is to concentrate on problems of data--flow and data reduction, which directly affect the deadtime or even the feasibility of the experiment. Coping with these problems already requires a substantial effort in programming intelligent devices.
In addition, we need to keep in mind that in a modern online system a huge effort goes into monitoring, control, and the organization and presentation of information. A typical large experiment may have hundreds of thousands of lines of code for detector monitoring and calibration, most of it detector dependent and written by experimental physicists for this experiment. A roughly equal amount of code provides a framework for the experiment--dependent programs. This latter code is often general, and may be used by different experiments. It complements the facilities, such as operating systems or compilers, which are provided by manufacturers. Together, the general (but HEP--produced) code and the manufacturer's software form an environment in which the experiment--specific applications are developed and run. The quality of both the application development environment and the final running system are crucial.
The following list of items gives some of the background to data acquisition systems currently in use or under development.
At the moment, the market for new recording devices/technologies is exploding, and we recommend some form of CERN-- or HEP--wide coordination to rationalize the situation. There is a need to standardize the recording formats as well as the types of media in order to support the transport of data between the many online and offline systems used in a typical experiment.
This working group has not attempted to reduce the difficult problems described above to a simple list of recommendations. What we offer instead is a summary of some of the important areas, with indications of future directions. Before entering these details, it is worth noting that, while the raw performance required for LHC is rather awesome (bytes/sec, channels, decisions/sec, etc.), the real difficulties are likely to lie in the extremely intricate collection of software required to implement such a system. Such problems of complexity are already encountered in the present generation of experiments, and will be even more serious for LHC. Apart from the challenge of actually making such a complex system work, we need to make sure that it presents a coherent and comprehensible interface to the physicists who must operate the experiment.
There will be many thousands of powerful processors embedded in the experiment and appearing in arrays for higher level processing. It will require an unprecedented level of organization and discipline to create and maintain such a system. The design process must be more formal and sophisticated than anything attempted before, and must use the best computer aided design tools which are available. It will be necessary to spend a substantial amount of time defining, simulating, and optimizing the system at the architectural/protocol level before proceeding to the detailed implementation phase. In addition, the very high data flow rates will require very careful integration of the detector digitization electronics with data compression and trigger processing, thus coupling the design of the detectors and trigger/data acquisition system from the very beginning.
A more structured, 'top--down' design is vital for a system of this complexity, but it is especially difficult to carry out in the relatively unstructured and multi--institutional environment of high--energy physics. The use of modern design tools allows a group of designers (be they physicists, programmers, or engineers) to describe a design at the behavioral level, in terms of large blocks, and then proceed to specify the internal structure and detailed timing of each component. As more timing information is added, the corresponding simulations of the system become more accurate, allowing further refinements in the overall design. One can imagine that these kinds of detailed simulations of hardware and software will play as significant a role in the design of the data acquisition and trigger for an LHC experiment as the use of Monte Carlo simulations does today in the overall design and understanding of a detector. As part of this process, clear specifications are created for interfaces between the components of the system which are independent of the details of the implementation, allowing the distribution of design effort without losing control of the coherence of the system. This structured design process has been found to be essential for the commercial design of complex systems, and HEP has a great deal to learn in this area before embarking on the design of an LHC experiment. " We therefore recommend to support pilot projects of combined software and hardware engineering methodologies, in order to select or develop methods and tools that can be used to design the next generation of data acquisition systems. "
Application Specific Integrated Circuits (ASIC) and VLSI Methods
The large channel counts and high performance demanded from the front--end electronics
(pre--amplifiers and digitization circuits) needed for a typical LHC experiment will
require extensive use of VLSI methods, with the additional complication of a high
radiation environment. It may also be necessary to include actual processing elements in
these front--end circuits to support the management of the large pipeline required for the
LEVEL1 trigger, as well as the data compression required (it may not be desirable to
transport the vast quantities of data generated by the detector elements). This is an
extremely complex area, involving the overlap of the detector design and the first levels
of the trigger and data acquisition system. It is clearly an area requiring much more
detailed discussion, such as that at the Fast Trigger meeting held at CERN in November
1988. " Given the need for very sophisticated design skills, it is strongly
recommended that CERN invest significantly greater resources in the application of VLSI
techniques to trigger and data acquisition problems than it has in the recent past. "
We would think that one of the most promising ways to acquire new skills and knowledge,
which has proven very useful in the past, is the pilot project. This allows one to gain
experience with new tools and techniques, while solving a useful problem and avoiding the
time pressure which often forces the use of conservative solutions.
In this area, we try to summarize the situation with VMEbus and Fastbus, and then attempt to gaze into the future to see what is coming next.
VMEbus The major advantage of this bus is the commercial availability of a wide
selection of processor and peripheral modules. There were initial difficulties because of
the inadequate level of standardization - modules from different manufacturers didn't
always coexist happily, but these have largely disappeared with time. The mechanics and
cooling are more appropriate for single crate desktop systems, and the small board size
makes it unattractive for front--end electronics. There are also difficulties because of
the lack of multi--crate and multi--processor standards and concepts, i.e., the error
handling and arbitration are awkward, there are no standard registers, configuration is
generally by jumper instead of software, the interrupt support is primitive, there are no
geographical addresses, and each large system must invent its own interconnect hardware.
These problems make it very difficult to create any standard software for a
multi--processor system. Nevertheless, despite its many technical limitations, VMEbus has
been extremely successful because the large commercial market results in reduced costs and
immediate availability of the latest technology, as well as the availability of commercial
software. This availability means that upgrades in CPU power or memory density can be made
fairly easily.
On the other hand, the use of Fastbus in front--end electronics has been fairly successful. The large board size, mechanics and cooling have contributed to the implementation of high density, economical systems. Furthermore, the standard includes high speed synchronous transfers, and the terminated ECL backplane of Fastbus is not yet close to its bandwidth limitations, leaving more room for future expansion, and perhaps providing a better match to the needs of the next generation of experiments at a hadron collider.
The conclusion is that these two busses have somewhat complementary advantages, and are both likely to survive in the future. In current designs, there is a tendency to use Fastbus at the lower levels in the system (industry rarely makes the digitization modules that are needed, either in Fastbus or VMEbus, and Fastbus has many advantages here), with VMEbus at the higher, processor--oriented levels (where the commercial support is substantially better in VMEbus).
It is quite likely that new bus systems will emerge in the commercial world before the
construction of any LHC experiment occurs. The success of VMEbus is a strong indication
that the next bus system used in high energy physics should be an industry standard, but
it seems unlikely that industry will find it important to implement large multi--crate
systems with a distributed high bandwidth interconnect. Nevertheless, there are two
short--term solutions that bear watching. VXI is a proposed extension to the VMEbus
standard, oriented towards instrumentation, with larger boards, better cooling, and ECL
support among other features. It should make VMEbus a much better place to build
front--end electronics. Futurebus is a proposal incorporating much better support for
multiple processor systems in a standard which could be strongly supported commercially,
and provide a migration path from both VMEbus and Fastbus. For the longer term, there is
the Superbus proposal (now known as SCI, for Scaleable Coherent Interconnect) which is a
very high speed, transaction oriented, point to point system benefitting from advances in
high speed network protocols. This could be of interest as the framework for building very
powerful processor arrays.
Everyone agrees that we should continue to follow emerging standards. Simple and
relatively cheap hardware interfaces are available, but the integration of standard
communications software into the environment of front--end processors is still difficult,
and further improvements are certainly desirable. The use of network software to integrate
the many microprocessors buried in the data acquisition system with the host computer is a
prominent feature of many of the major present data acquisition systems at CERN and DESY.
In general, the coherent integration of standard network software into the system software
running on the microprocessors is a very important requirement for any microprocessor
software packages to be considered in the future. Such systems should support a standard
transport level, as many packages do now, but more complex layers of software are also
required. Standard examples include remote file access (including directory modifications)
and remote login. As a further example, online networking tends to be very transaction
oriented, and high level packages such as Remote Procedure Call (RPC) are proving to be
very useful for distributing functions within a multi--processor system. Languages Data acquisition systems contain a mixture of time critical problems and high level
problems, making it unlikely that a single language can provide the optimum solution for
all problems. It is important for the future that we be able to use modern languages which
are structured and well supported by software engineering tools. It is unlikely that
FORTRAN will meet these requirements (although its use in some online applications is
likely to be favoured by physicists for some time). There is already a strong trend to use
the C language in the data acquisition environment. C is fast gaining importance in the
outside world, as a reasonably portable and efficient systems programming language. Its
links with the Unix operating system are another factor in its growing popularity. The computer science community has placed significant emphasis on designing languages
for large, multi--processor software projects. It is possible that we could benefit from
their efforts, which have produced languages like ADA - such languages can encourage more
modular, portable software and may support a measure of object--oriented design. ADA also
has associated with it a powerful development environment. In addition, these languages
can simplify programming in a multi--processor system because they already contain
primitive concepts which are usually the domain of the operating system. An example of an interesting combination of language and hardware is the INMOS
Transputer supporting the OCCAM language. OCCAM is particularly powerful in its ability to
allow users to develop multi--processor software on a single processor and then distribute
the applications over many processors (although this language is not readily available for
other hardware). Taking a somewhat more futuristic approach, one can hope that we will enter an era
where the use of behavioral descriptions and very high level specifications will start to
supersede the actual writing of software in the current sense. Unfortunately, this
objective is difficult and it may not even arrive in time for LHC. We conclude that
physicists need much more experience with the use of modern languages, and in particular
in their application to large software projects. It is desirable to provide a coherent software environment, allowing users to work with
a complete set of similar tools at all levels, from the Online Computer down through the
general purpose processors to the application specific processors. This coherent
environment will need to include support for centralized databases, resource management,
run control, error and status reporting, information presentation, human interfacing, and
all of the other services required for the organized collection of data. It should be
integrated as far as possible with the corresponding facilities of the computer centre or
distributed offline processing and interactive facilities. At present, Unix seems to be the only candidate for a hardware--independent
industry--supported operating system. It is already available on many workstations and
widely used as a basis for microprocessor cross--software development systems. It is
offered by DEC on the VAX family of computers, and is on its way to becoming more widely
available on mainframes as well. Many efforts are underway to improve the current level of
standardization, most notably the Open System Foundation (a consortium of many industry
giants) which is committed to creating a non--proprietary standard. However, Unix is not uncontroversial. The VMS operating system is very popular, and has
been chosen by the vast majority of present experiments in the online environment. This is
probably due to its scope, user--friendliness, and manufacturer support. Second, Unix is
known to have real--time problems, which may make it unsuitable for many applications on
embedded microprocessors, without kernel modifications. There are efforts to make a
'Real--Time Unix', but it is not yet clear how successful this will be or how it will
relate to attempts to standardize real--time kernels. Independent of any arguments about
the merits of Unix, no experiment known to us today uses Unix in the online environment. It is therefore by no means clear what will be the future role of Unix in the online
environment. Nevertheless, in view of its potential in providing an integrated environment
on a wide range of computers, the working group feels that Unix must be taken seriously as
an option for the future. " We recommend that a study should be made of whether
and how to introduce Unix in the data acquisition environment. This study should take into
account the balance of technical and commercial arguments for Unix vis--a--vis other
systems, the opinions of users, and the extent to which Unix can be used in real--time
applications. " The area of real--time kernels is finally coming under scrutiny
by a sub--committee of VITA (VMEbus International Trade Association), which contains
representatives from all major producers of real--time kernels. Work is proceeding on
calling conventions, driver interfaces, and the harder problem of a standard real--time
kernel definition. The latter is known as ORKID (Open Real--Time Kernel Interface
Definition). It should allow vendors to provide standard drivers with their peripheral
boards, etc. (At present this work is oriented toward the Motorola 68K family of
processors). Nevertheless, in this area there seems to be a conflict between the real--time needs
for front--end processors and the full operating system features which many find
desirable. There are essentially two schools of thought, although one can hope that this
situation will disappear in the future with the release of a new system combining the best
features of both worlds. The advantages of the first approach are a uniform high--level software development
environment on a workstation with windows and graphics, combined with full real--time
performance in the front--end processor. The disadvantage is the requirement of sufficient
high--level CPU power to support fast response during software development. On the other
side, a system which provides a high quality set of compilers, editor, and multi--task
debugger in a friendly way requires a complex hardware/software environment. It is not
clear that we really want all of the complications of a full operating system in our
front--end processors; there may also be penalties to pay in performance. " We
recommend that standardization efforts on real--time kernels, and associated system
software should be followed closely, and taken into account in planning future data
acquisition systems. " Distributed Online Systems A key technical question is: how can the tasks of an online
system best be distributed to achieve efficiency and cost--effectiveness, whilst
maintaining an effective user--friendly control, reporting, and information management
system? There is no easy answer to this: it needs serious study based on experience with
existing systems and the availability of modern techniques and tools. There are at least
two kinds of distribution to be considered: The distribution of the functions performed up to now by the online computer, their
modularization by sub--detector (with later integration into the overall production
system), and the incorporation of "off--line" style tasks into the online
environment, have usually been tackled in an ad hoc way. More formal techniques for distributing functions will need to be studied, adapted, or
developed. Some are more appropriate to loosely--coupled systems and others to
tightly--coupled ones. They include facilities associated with languages such as ADA or
OCCAM, techniques such as Remote Procedure Call (already widely used at CERN in online
applications), object--oriented systems, distributed operating systems and many others.
" The question of how best to distribute the tasks of an online system in an
efficient and cost--effective way should be investigated. HEP needs to participate more
actively in following the trends in research, and associated industry developments. "
In addition, although functions may be distributed, it is essential to give a coherent
picture of the experiment to the physicists. This requires attention to the management of
information in the distributed environment and ease of access to it. Debugging in a
distributed environment is also very important. " Both database--management
techniques and user--friendly human interfaces are examples of areas where development is
rapid outside our field. We need to keep abreast of these developments and learn to apply
them in our environment, complementing them with in--house developments where necessary.
" This whole subject is the domain of a separate group, but its significance, especially
for LHC can hardly be over--emphasized. It is important to note that these tools serve a
role both on the technical level, reducing design errors and optimizing performance, and
also on the social level, by allowing the coherent development of a complex design by many
autonomous groups. Given these arguments, general--purpose packages are likely to be an important
component of future data acquisition systems. Their planning, and combining their creation
and maintenance with formal design methodologies, will be a central issue of future
support. Reliability The issue of reliability will take on much greater significance in the complex systems
which we are proposing. It is associated with other issues such as error/exception
detection and handling, fault tolerance and redundancy. Today's systems have been designed
with very little consideration for these issues, and they survive because of their
relative simplicity. A simple example is the transport of data in a contemporary data
acquisition system. This data works its way through a series of memories, which generally
have at best a parity check, and then proceeds via cable segments, which again contain at
best a parity check, finally arriving at the 'Online Computer'. In the end, there are many
possible failures which can remain undetected for long periods, requiring frequent
hardware testing in order to stumble across the problem. This method of operating is
inconsistent with the continuous use of the data acquisition system which will be required
by the experiment itself. In the future, it will probably be necessary for individual processing elements to be
able to detect hardware faults and also for the system as a whole to isolate and tolerate
faults of many kinds - if not, the mean time between failures is likely to compromise the
operation of the system, as well as the user's confidence in the final results. Although everyone expresses some fear that too much standardization can prevent the
kind of innovation we need to keep up with technological advances, many reasons can be
cited in favour of standardization: This is a very complex area, with many possible levels of standardization :
international (IEEE, ISO, ESONE, etc.), groups of manufacturers (original case for VMEbus,
3480 cassette, etc.), and HEP (HBOOK, ZEBRA, etc...). A general conclusion of our
discussions was that HEP should try to define areas where its needs are similar to those
of industry, in which case we should try to standardize and use commercial products as
often as possible (be they hardware or software), as opposed to those areas where we will
continue to require unusual products, in which case HEP should try to reap the benefits of
the revolution in computer aided design tools to ease the job of building our own
products. " A major factor in evaluating whether to use a commercial versus local
product should be a full cost estimate, including labor costs as well as proper estimates
of the true maintenance requirements (especially for software). " Some areas where standardization is interesting are : Industry is making progress in many of these areas: we have pointed out some recent
developments in an earlier section (Languages and Operating Systems). This issue of
standardization is a general one and applies to other areas as much as to the online
field. Careful consultation with users is required as well as attention to technical,
support and commercial factors. The size and loose structure of modern collaborations make
solutions quite difficult in our community. " We need to set up mechanisms to make
sure that the user community participates in the selection and/or definition of standards,
so that the standards selected are widely used and properly supported. " Several general comments can be made, in addition to the points raised in the
Standardization section. Medium--term evolution of data acquisition systems In view of the arguments above, we
need to plan for the evolution of data acquisition software both for conventional
experiments and in preparation for LHC. We should develop a model of what the data
acquisition system of a "conventional" experiment starting around 1991 should
look like. An analysis of what is needed for such an experiment, and what is available
both within CERN and outside, would allow us to make more serious estimates of what work
is required. This study should be a joint effort between experienced users and online
support and development staff. Such a study could be linked to the startup of a real
experiment. It would be complementary to the research required for the special needs of
LHC; liaison between these two activities should be as close as possible. It is also clear from the preceding sections that it will be important to embark on
research and development work in many areas in order to create the hardware and software
for an LHC trigger and data acquisition system. Some of this work is already starting in
the form of workshops such as the 'Meeting on Fast Triggers, Silicon Detectors, and VLSI'
held at CERN in November 1988, and the 'Workshop on Triggering and Data Acquisition for
Experiments at the SSC' (Toronto, January 1989). Their main topic is the close
relationship between detectors and the trigger and data acquisition system as necessary
for LHC, as well as the overall system architectures which might be needed. We should emphasize here that 'R&D' mostly relates to technologies or methods
already used elsewhere, but largely unfamiliar to our community. Such R&D may be to a
large extent a question of learning how to use, and adapt where necessary, the more
sophisticated tools and products of the commercial world. For instance, an R&D project
could involve learning to use structured design and simulation tools to create a complex
processor module and to optimize its design inside a particular model for a data
acquisition system. This does not involve true research, it is more a question of
education in the use of certain tools and techniques, and an adaptation of those tools
(and our way of thinking) in order to solve the difficult problems posed by the LHC
environment. We emphasize the need to modernize the approach of physicists, engineers and
programmers to solving problems in the field of data acquisition. This process is most
effective if it occurs while solving realistic problems, but it should take place in an
atmosphere which is somewhat free of the need to meet the deadlines which govern currently
running experiments. Only in this sense we call it 'research' as opposed to 'support'.
" This work will need longer--term support, and we strongly recommend that CERN
trigger a discussion on the means for supporting long--term projects in the general area
of trigger and data acquisition system development. It seems clear that CERN cannot tackle
this problem alone, and that a broader European initiative will be needed. "
Central and Distributed Processing and Related Questions
Richard Mount | (Caltech/L3, convener) | Brian Carpenter | (CERN DD/CS) |
Manuel Delfino | (Barcelona/Aleph) | Fridolin Dittus | (Caltech) |
David Foster | (CERN DD/SW) | Chris Jones | (CERN DD) |
Hans Hoffman | (DESY) | George Kalmus | (RAL) |
Mike Metcalf | (CERN/DD) | Alan Norton | (CERN/UA1) |
Dieter Notz | (DESY) | Alan Poppleton | (CERN/UA2) |
Andre Rouge | (IN2P3) | Emilio Pagiola | (CERN/EP) |
Les Robertson | (CERN/DD) | Julius Zoll | (CERN/EP) |
Usage of Data Bases in Experiments Luc Pape (CERN/Delphi, convener) Ken Knudsen (CERN/Omega) Andy Parker (CERN/UA2) Alois Putzer (Heidelberg/Aleph) Harry Renshall (CERN/DD) - - Software Engineering for Experiments Tony Osborne (CERN/DD, convener) Nick Ellis (CERN/UA1) Stephen Fisher (RAL/Aleph) Fred James (CERN/DD) Gottfried Kellner (CERN/Aleph) Henri Kowalski (DESY/ZEUS) Karl Gather (DESY) Chris Onions (CERN/DD) Paolo Palazzi (CERN/Aleph) John Poole (CERN/LEP) Otto Schaile (CERN/Opal) Roman Tirler (Saclay) Ian Wilkie (CERN/SPS) - - Graphics Support Francois Etienne (IN2P3--CCPM, convener) Daniel Bertrand (Brussels/Delphi) Rene Brun (CERN/DD) Alan Grant (CERN/Delphi) Frank Harris (Oxford) David Myers (CERN/DD) Jean--Pierre Vialle (IN2P3--LAPP) Wojciech Wojcik (IN2P3--CC) Training Stephe O'Neale (CERN/Opal, convener) M.Goossens (CERN/DD)
A study of off-line computing and general software support for the CERN physics programme in the 1990s. Including:
This chapter will examine the needs of each area of the CERN physics programme at the varying levels of detail allowed by existing studies and the plans of individual experiments. The total needs will be summarized in the following chapter, together with a more general consideration of the distributed computing environment which can best meet these needs.
The computing needs of the LEP experiments have been reviewed in some detail in the MUSCLE [Footnote: The Computing Needs of the LEP Experiments, CERN/DD/88/1, January 15, 1988. ] report. The estimates were made per experiment, on a global (worldwide) basis, for the "years" "1989", "1990" and "1991" when the number of Z events accumulated by each experiment reaches 1 million, 4 million and 10 million respectively. [Footnote: On the assumption that these "years" will have 3000 hours of beam time, the corresponding average data taking rate for Z events during beam periods is 330 events/hour in "1989", 1000 events/hour in "1990", and 2000 events/hour in "1991". ] The following three sections are reproduced verbatim from the MUSCLE report.
Processor Power The following table summarizes the estimates for the different types of processor power that will be needed by each experiment. The attentive reader will notice the question mark attached to the Total? line. It is, indeed, rather unclear (and perhaps not even very meaningful to ask) how the different types of processor power should be added in order to come to a total. This is because there are many ways in which this power can be provided; some options permit sharing the power between different tasks, while for other options such sharing is effectively excluded. The major sharing that we believe might be effective would be to carry out some of the simulation work either using the interactive capacity (workstations etc.) during the middle of the night, or using the capacity needed to generate the master DST outside beam periods.
"Year" "1989" "1990" "1991" ------------------------------------------------------ Monte Carlo generation 4 12 24 Processing of MC events 1 3 6 Generating the master DST 3 8 16 Accessing the DSTs 8 9 13 Extracting the physics 8 8 8 ------------------------------------------------------ Total? 24 40 67 ------------------------------------------------------ Processor Power per Experiment (CERN units) Table 1
Data Storage The following table summarizes the estimate for the total volume of data, in Gigabytes, that will have to be stored by each experiment. Note that the expansion needed for making duplicate copies is left to the appreciation of the reader. By the end of 1991, if the data are held on cartridges of 200 Megabyte capacity and no cartridge can be recycled, then each experiment will need a total of at least 35000 cartridges to hold these data.
"Year" "1989" "1990" "1991" ------------------------------------------------------ Raw data 400 1600 4000 Master DST (real events) 20 80 200 Team DSTs 40 200 440 Personal DSTs 10 10 10 Simulated raw data 200 800 2000 Simulated DST events 20 80 200 Duplicate copies ? ? ? ------------------------------------------------------ Total 690 2770 6850 ------------------------------------------------------ Accumulated Data Volume per Experiment (Gigabytes) Table 2
Data Manipulation
"Year" "1989" "1990" "1991" ----------------------------------------------------- With no disk space 40 50 65 With 100 Gigabytes disk space 2 10 30 ----------------------------------------------------- Cartridge Mounts per Hour per Experiment Table 3
The MUSCLE report restricted its scope to the first 30 months of LEP operation. What
happens beyond "1991" will depend on the physics uncovered by LEP in the Z
region, and by the progress in upgrading the machine to LEP--200. A major unknown is
whether a programme of physics with polarized beams will become part of the plan for the
exploitation of LEP. To give an indication of the LEP computing load beyond
"1991" it has been assumed that the installation of LEP--200 cavities starts in
1992 and continues beyond 1995. It has also been assumed that energy scans above the Z are
interspersed with running on the Z. Even if polarized beams are not available, additional
Z running is likely to be desirable to take full advantages of detector improvements made
during the initial running; in some cases major detector components are scheduled for
installation as late as 1991. If LEP were to stop suddenly at the end of 1991, physics
analysis would continue, initially needing most of the resources required during running,
and reducing to minimal needs by about 1995. In the more likely scenario outlined above,
the fall in Z--related needs would be more gradual, and is very approximately estimated in
The CERN SPS Collider programme is expected to remain active well into the 1990s. The associated computing requirements may be estimated by extrapolation from present experience, taking account of machine and detector upgrades and of the trends of the physics programme. This extrapolation also gives an indication of the likely scenario in the LHC era. It is a fact that hadron collider experiments are limited by the available on--line and off--line computing resources. The on--line limitation comes from a dramatic mismatch between the total interaction rate (50K events per second at a luminosity of 1030cm-sec-1) ) and the rate at which events can be written onto tape (typically five 200 kilobyte events per second with present technology). The goal of the data acquisition is to ensure that only "interesting" events are recorded. This is partly achieved by triggering on the presence of e.g. a high p lepton, missing energy and/or jets. Such triggers alone are adequate for W events, but not for more complicated topologies e.g. heavy flavor events where increasingly sophisticated higher order triggers, basically on--line processors, are used to reduce the rate in successive steps. At expected ACOL luminosities (3x1030cm-2sec-1), UA1 will use at least 12 units of on--line computer power in a parallel processing configuration (3081E emulators) as the final trigger level. On--line event rejection based on complex algorithms and uncertain calibrations is of course risky. Given faster data acquisition and sufficient off--line power, a safer procedure would be to write more events and be more selective off--line. For high cross--section processes this would also allow more useful events to be collected; for example, at a luminosity of 3x1030, the rate of b&obar.b events is about 30 events per second (compared to one event every 5 minutes for W production)! For off--line processing, the ultimate goal is to provide sufficient power to keep pace with the saturation capacity of the data acquisition. We assume that 50% of all events will be passed through something approaching full reconstruction; given detector and triggering improvements, events will either be usable for physics or will come from a background source which cannot be easily rejected at an early level. With the present data acquisition technology, this means that 13 million events could be written in an FME (Full Month Equivalent = one month of running at 100% efficiency); these events would occupy 8500 cassettes of the currently available 0.2 Gigabyte capacity. Taking UA1 as an example we can anticipate processing times of the order 30 seconds per event (standard CERN accounting seconds), and therefore about 7 CERN units would be needed to process the full FME (one CERN unit assumed to be 7000 hours/year). In a calendar year, a single experiment may expect to get two FMEs from four months of running time. Furthermore the first pass through the data should match the data acquisition speed. The requirement is therefore about 15 units, preferably in the form of a dedicated 45 unit processing facility available over four months. We can expect that further data analysis and related Monte Carlo running will require at least twice the time for the first data pass, i.e. about 30 units. The Monte Carlo running can conveniently be run on dedicated facilities, while the remainder, say 15 units, should be in the form of general services. CPU power is not the only parameter in the present scenario. The logistics of handling large volumes of data are already apparent today and are likely to become the limiting factor if denser storage media are not introduced, both in experiments and in computer centres. The above scenario implies 17,000 raw data cassettes per year, giving at least 34,000 total to cover Monte Carlo outputs, DSTs etc, again for a single experiment. An increase in density of at least a factor ten is highly desirable to ensure that the CPU power is accompanied by the appropriate ease of use. UA1 has undergone a major upgrade to the muon detectors and to the data acquisition system. In the near future a sophisticated uranium--TMP calorimeter will be installed. The UA1 computing load can therefore be expected to rise from about 30 units in 1990 to about 60 units by 1992. No further upgrade is foreseen for the UA2 detector, and the throughput will be limited by the present data acquisition capacity. The UA2 requirements are therefore not expected to exceed 10 units over the period 1990 to 1992. Qualitatively we may expect that the requirements of an LHC experiment will be similar to those discussed here for the SPS Collider. The on--line triggering will require a significant upgrade in order to operate in a many bunch environment with as little as 5 nanoseconds inter--bunch gap. The LHC will depend on high luminosities in order to reach new discovery thresholds, and 1033cm-2sec-1 is now discussed as "standard" with 5 x 1034cm-2sec-1 in a special high luminosity intersection. The latter figure implies about 1010 interactions per second with, in the mean, 25 overlapping interactions per "event". Initially it is likely that highly selective triggers will be used, for example four muons from (hopefully!) the decay H @A ZZ, producing a relatively low rate of especially clean events from low cross--section processes. As with the SPS Collider this phase may then be followed by studies of more complex topologies and higher cross--section processes.
There have been no systematic studies, comparable to the MUSCLE Report, of the needs of the other CERN experiments. A simple working hypothesis, that non--LEP computing needs would be similar to the total LEP computing needs, has been in use at CERN to make approximate predictions of the proportion of CERN's resources which would become available to LEP experiments. It is probable that if CERN continues to support programmes of non--LEP, non--Collider physics, the experiments will also generate more and more data requiring more and more CPU cycles. More sophisticated electronics and computing techniques will be employed to increase the potential of detector hardware in an evolution typified by the pattern--unit to ADC/TDC to flash--ADC progression of recent years. This evolution is stimulated by the continuing increases in the cost--effectiveness of data--acquisition electronics. Since the same technologies also improve the cost--effectiveness of off--line computing, the fractional cost of computing (with respect to a complete experiment) need only increase very slowly, if at all.
More than 25 LEAR experiments have now been approved involving over 500 physicists. The largest collaboration is CPLEAR with about 90 physicists. CPLEAR has made very preliminary estimates of their data volumes and CPU needs. They expect to take 2.5 x 10 good events in three years. With an event size of 2 kilobytes and guessing an equal volume of background, CPLEAR will accumulate 10,000 raw data tapes. The master DST volume is expected to be comparable. The necessary processing power is estimated to be 9 unit--years, assuming only one pass through the data. Extrapolation from these figures to the total LEAR need is not easy, but it is probably safe to say that the total LEAR data handling requirement will be similar to that of a LEP experiment, while the CPU requirement may be somewhat smaller than that for a LEP experiment.
There are currently more than ten SPS experiments taking data or in preparation. As in the LEAR programme several hundred physicists are involved. Although the long term future of SPS fixed--target physics is unclear, strong physics interest remains in making high statistics measurements to yield, for example, nucleon structure functions or the parameters of CP violation. Some collaborations, for example NMC, are preparing proposals for new experiments to last beyond 1992. Typical current use of computing power by an SPS experiment is in the range 1 to 6 units, with some experiments expressing the view that they could make profitable use of much greater computing resources. Data volumes of several thousand tapes per year are also common. For example NMC wrote about 11,000 tapes in the last year, of which one third were raw data.
In addition to LEAR and the SPS programme, the SC programme and the many detector development and calibration activities cannot be ignored. As an example of the latter, the calibration of the L3 electromagnetic calorimeter has consumed 2 units for most of the last six months. A further indication of the current needs of the non--LEP, non--Collider physics can be obtained by observing the recent use of the CERNVM service. The 'other experiments' now consume much more than half the total service. Most of this consumption is achieved in the face of considerable turnaround penalties (for exceeding the target allocation), showing that the physicists concerned consider that they have a real computing need. In the absence of a systematic study, there can be no attempt to use the examples above to obtain quantitative estimates for the future. However, it seems likely that the 'other experiments' will not require less resources than one LEP experiment. At the other extreme, it seems very unlikely that their need would be greater than that of three LEP experiments.
HERA Experiments do not form an explicit part of the present study. However, it has
been considered valuable to include a summary of HERA needs for comparison with the CERN
estimates. The data length for raw events is of the order of 100 kilobytes and for master
DST--events 20 kilobytes. We assume on average 1 event/second and 0.5 years of Hera
operation per year. This gives 1.5 x 10 raw events/ year/experiment and 1.5 x 10 master
DST--events/year/experiment with Q > 3 GeV. The total amount of data per year and
experiment is 1500 Gigabytes for raw events and 30 Gigabytes for master DST--events. The
computer time for reconstruction is assumed to be 20 seconds (on IBM 370/168) giving 1.5 x
10 x 20 computer seconds per year or 30 x 10 = 12 Units/ experiment (290 days/year = 2.5 x
10 sec). For Monte Carlo event generation we assume to produce one simulated event for
each master DST--event = 1.5 x 10 Monte Carlo events/year/experiment. 30% of Monte Carlo
events have a full length of 100 kilobyte/event and 70% have a DST length of 20
kilobytes/event. For data storage one requires on average 44 kilobytes/event or 1.5 x 10 x
44 kilobytes = 66 Gigabytes/year/experiment. The Computing time ranges from 60 to 900
seconds for 50% of generated events, varying from small to high Q and from 5 to 10 seconds
for 50% fast Monte Carlo events. This gives in total 1.5 x 10 events x 60 seconds = 9 x 10
computer seconds per year or 3.6 units. 0.4 units are needed for reconstruction of Monte
Carlo events. Assuming that Monte-Carlo generation and testing is performed 2.5 times,
this would add up to 10 units. To access DSTs and to extract physics we assume a CPU need
for one hour/day/physicist. For 200 physicists/experiment this would give 200 hours/day or
10 Units (= 200 hours/24 hours/0.85 efficiency). In Summary: Each Hera experiment needs 36
Units CPU power and 1600 Gigabytes of data storage. At Desy it is planned to install 15
Units CPU Power per Hera experiment plus 15 Units for infrastructure. The needs for Hera
experiments in terms of CPU and data storage are lower than those for LEP experiments.
This comes mainly from two facts:
We assume that in the future, as is the case now, HEP experiments will be undertaken by
teams, usually comprising a large group of university physicists with smaller
contributions from national laboratory and CERN physicists. Although this report does not
concern itself on how HEP experiments should be done, the balance of the intellectual,
technical and financial contributions of universities, regional Laboratories and CERN does
have an impact on the overall organization of computing. This mode of operating, with
large and often geographically scattered, collaborations may not necessarily be the most
efficient way to conduct any single experiment. However, this division of effort has great
benefits; it ensures the continued inflow of talented young physicists throughout Europe
and has the additional value of encouraging independent thought and lines of communication
which may be stifled or discouraged if most resources are directly controlled by a single
management. Many of today's detectors, including the most complex, are designed,
constructed and the data analyzed by university and national laboratory groups at their
home institutes. However, several forces conspire to drag all but those with heavy
teaching loads or little travel money to CERN. These forces should be minimized so that
the physics largely emerges in the institutes rather than centrally at CERN. The forces
are:
Forces (1) and (2) are not the concern of this report. Forces (3) and (4) should be
neutralized as far as possible by improving electronic interpersonal communications, and
by improving access to computing facilities and data from home institutes. In the last ten years about half of the data processing and analysis of CERN
experiments has been carried out at regional centres and at universities. An incentive to
"export" physics analysis to the collaborating institutes was provided by the
"one third/two thirds" rule under which CERN experiments were expected to find
two thirds of their computing needs outside CERN. In practice, this has not often been
achieved, the average being closer to one half. The 2:1 ratio was evaluated in terms of
standardized CPU units. When the majority of computing was done on mainframes, these units
were a reasonable basis of comparison. However, now mainframes, specialized processors,
mass storage, large numbers of workstations and networking are all important and expensive
components of the HEP computing environment. It therefore no longer makes sense to express
support of physics analysis in terms of a simple CPU based rule. Nevertheless, considering
that 1/2 to 2/3 of the physicists actively working on experiments do not need to be at
CERN, and that somewhat greater resources are needed to support geographically dispersed
physicists, it is reasonable to continue to aim to provide about 2/3 of the total
computing resources outside of CERN. Every effort should be made to provide non--CERN
based physicists with similar access to computing facilities and the experimental data to
that available on the CERN site. National/Regional computing centres should play an important part in providing for the
total computing requirements of the CERN community. This is important for many reasons:
In the past these centres have played an important role in the overall HEP computing
scene in Europe and there is every reason to expect this role to continue into the future.
For this to be efficiently carried out good networking between the regional centres and
both CERN and the local university community is a necessity, as is a mechanism for the
exchange of bulk data between CERN and the centres. A full exploitation of the technically
possible 'European HEP analysis environment' with increasingly automated exchange of data
between CERN, regional centres, and universities, will require an unprecedented level of
cooperation between all parties, in which the role of regional centres will be
particularly important. More quantitative recommendations on the evolution of regional
centres are given in section Private facilities are considered to be systems owned by a single group, or by a single
experiment, which are large enough to provide a significant fraction of an experiment's
computing needs. It is recommended that the establishment of private general purpose
computer centres at CERN be discouraged. If the facilities of the CERN computer centre are
adequate, and the policy for their future development is firmly founded on the concept of
allocating HEP's resources to optimize physics output, it makes no sense for individual
experiments to set up and run general purpose computing centres at CERN. The case of more
specialized facilities is less clear. This report is not concerned with on--line event
filters, but the processor farms which perform the filtering may also make excellent
facilities for some of an experiment's production processing. Purely off--line private
processor farms are being used effectively by several experiments, and it is expected that
such facilities will continue to play a valuable role both on and off the CERN site. To establish a reasonable working model, it is necessary to consider both the
technology and the requirements. In this respect data volumes play a large role in
deciding where CPU operations should take place. Each LEP experiment will generate around
10,000 (200 Megabyte) raw data tapes in a year of running on the Z. Volumes of simulated
data are likely to be comparable. Planned volumes of the 'Master DST' (reconstruction
program output) range from about 20% of the raw data to more than 100% of the raw data. If
the Master DST is small, it is assumed that any detailed study of an event will require
access to the raw data tapes. The high volume Master DST's will contain all the
information likely to be needed for detailed analysis. In either case, physicists have a
potential need for random access to data stored on tens of thousands of tapes. Clearly,
the ability to handle such volumes of data is a resource which is at least as valuable as
CPU power, and which must be regulated and allocated with at least as much care. To reduce
these data to the volume which can be handled by individual physicists and workstations (a
few tapes at most), will require a hierarchy of DST's with progressively more stringent
selections on either which events are present, or what data are present per event. It is
extremely likely that substantial processing will be performed during the selection
processes. Given these large data volumes, how can off--site processing be maximized
without introducing delays and difficulties in physics analysis? A possible approach is:
Physicists are accustomed to the difficulties of exchanging data between computers with
different data representations. While these difficulties are never insurmountable, they
threaten to reduce seriously the transparency of a distributed data management system. The
HEP community should treat an IEEE data representation as a significant advantage when
considering future computer purchases. summarizes the evolution of the total batch CPU needs of CERN experiments up to 1992.
It has been assumed that in providing the UA1 need of 45 units during the four months of
data taking it is possible to make use of 50% of this capacity for other purposes when UA1
are not taking data. All the remaining non--LEP experiments, including other work at the
SPS collider have been assumed to be equivalent to approximately two LEP experiments,
which is an assumption allowed by the uncertainties described above. This category includes such machines as IBM, IBM--compatible, Cray (used on unmodified
code), and the larger VAXes. There is no doubt that almost all computing for HEP can be
performed on such mainframes, and, as explained later, the differences between the
(list)price/performance of the various machines is small. The large fraction of HEP
computing which involves high volume data handling, linked to substantial CPU needs, is
ideally suited for conventional mainframes. On most modern mainframes vector processing hardware is either built in, or a
relatively cheap optional extra. Thus only benefits can accrue from allowing the compiler
and the vector hardware to do whatever they can to speed up unmodified HEP code. However,
a substantial body of experience shows that the resultant increases in speed rarely exceed
10 to 20%. Although it is too early to present a final verdict on the success of
'vectorization' through code modifications, it is now widely accepted that it would be a
mistake to expect any substantial speed up for reconstruction and physics analysis code.
By its nature this code has to be flexible, and therefore unstable. Vectorization of this
code to achieve overall speed improvements of a factor two may well be possible, but at a
cost in manpower and code maintainability which most physicists would consider
unacceptable. There is considerably more optimism about the prospects for speeding up
Monte Carlo simulation through algorithm re--design rather than code modification. Work is
now in progress to produce a vectorized version of GEANT3 although it is too early to make
predictions of the benefits. It has already been demonstrated [Footnote: e.g. K. Miura
(Fujitsu USA), Lectures at the CERN Computing School, Oxford, 1988. ] that the stable,
experiment independent, components of a shower Monte Carlo can be re--written very
effectively for a particular vector machine, but up to now, no complete Monte Carlo
program has been vectorized. Vectorization of Monte Carlo code may make it possible to
generate simulated data at the level of detail, and with the statistical precision,
merited by current and future detectors. Continued efforts in this area should be strongly
encouraged. Most HEP computing exhibits inherent event--level parallelism. This parallelism offers
opportunities to control the cost and expand the flexibility of HEP computing. To define
the role of parallel computing for HEP in the 1990s, it is necessary to examine both the
fraction of HEP computing which can be parallelized, and to specify which of many possible
parallel architectures should be used. Most experiments estimate that 40 to 70% of their
minimum needs could be provided by parallel computing systems. Compared with the use of
conventional mainframes, significant but tolerable inconvenience is expected, which is
partially offset by the hope that cheap parallel systems could make it possible to exceed
the minimum CPU requirements, with corresponding benefits for physics. Exploiting HEP
parallelism by allowing one experiment to run a job using all the CPUs of an IBM 3090
brings little or no benefit. [Footnote: Some large programs may be required to use the
CRAY processors in parallel because they might leave insufficient memory for other
programs to exploit the remaining processors. This form of parallelism is not considered
further in this report. ] Benefits arise when the processors used in parallel are much
cheaper per unit of CPU power than mainframes. Price/performance can be improved by, for
example, using mass--produced CPU chips, or by restricting I/O and inter--processor
communications possibilities. Processor 'farms' have been used in HEP for some years.
Physicists who have replicated existing designs have found them cost--effective solutions
for stable 'production' programs involving limited I/O. Whether the various emulator and
'ACP' programmes have been cost--effective as a whole is more difficult to assess. A major
justification for the development of such processors has been their use as event filters
in data--acquisition systems. Thus many would argue that the development costs should be
ignored when evaluating the processors for off--line use. Whatever the true costs of their
development, these processors certainly allow a cost--effective expansion of CPU resources
for any experiment in which they are already heavily used. The optimum future parallel
architectures for HEP are not immediately obvious and this working group believes that the
CERN (or even HEP) community should initiate a continuing and coordinated parallel
computing project. The project should be rooted in reality, by requiring that it provide
almost immediately a parallel computing service to users, while exploring the most
promising approaches for the future. The most promising approach for early investigation
appears to be that of using commercially available processors adequately linked to the
data handling capabilities of a centre's mainframe system. The major challenge would then
be to make the parallel processing system as almost as easy to use as a mainframe of the
same total power. Other approaches, such as building special processor boards, or buying
complete parallel computers, should also be investigated. Central time--sharing services are still a cost--effective solution to the provision of
general--purpose interactive computing for an organization of CERN's size, [Footnote: A
paper of A.Osborne and L.M.Robertson, prepared in February 1987, calculated the average
cost of central computing, including all overheads (full costs of staff, management,
energy, buildings, etc.), as 137 SF per standard CERN CPU hour. This is approximately
double the cost calculated by taking into account only the cost of IBM hardware. Thus the
cost of the current VM service (supporting 450 simultaneous users at peak times, and 2,300
different users each week, consuming a total of 800 CPU hours) might be estimated as: 110
kSF per week; 250 SF per peak user per week; 50 SF per user per week. The cost of the basic hardware of a simple workstation today, including a minimal
contribution to the cost of a "boot--host" or disk server, is 15 kSF plus 25%
per annum maintenance. If we assume a three year lifetime for the station, this represents
only 170 SF per week per workstation. The hidden costs for workstations are almost
certainly higher than those of the Computer Centre, where economies of scale are possible,
especially if manpower--intensive functions like disk backup are included. No realistic
assessment of these costs has been carried out (to our knowledge), but it is clear that
the cost per peak user is very similar for both types of service. ] especially when
measured in terms of cost per individual user of the service. Measured in terms of cost
per active user at peak times, however, personal workstations of the VAXstation 2000 or
Apollo DN3000 class are very competitive, and in addition provide a quality of service to
the end--user which is incomparably better than that of a time--sharing service: human
interface with bit--mapped screen, windows, pop--up menus and mouse; guaranteed
availability - no delays, queues, interruptions outside the control of the user; personal
choice of important details of the system. The evolution towards personal workstations as
a vehicle for general--purpose interactive computing started early in 1987, following a
substantial improvement in price/performance over most of the market, and is now well
under way. However, there are a number of points which must be considered.
Personal workstations cannot be shared Personal workstations are personal - there must
be enough of them to ensure that a convenient one is available when required. The result
will inevitably be that these devices will often lie idle, but, as the above cost
comparison shows, we should not worry about this if the workstation is used regularly at
peak periods. If some batch work can be run on workstation clusters this should be treated
as an added bonus. The MUSCLE report made quantitative estimates of LEP data volumes, and assessed some of
the benefits conveyed by keeping some of the LEP DST data on disk storage. The duplication
of data needed for distribution to universities and regional centres was not included.
Wider discussion of the MUSCLE data volumes has tended to show them as lower limits,
particularly with respect to the 'Master DST's' which may be bigger than the MUSCLE
estimates by large factors. Noting that many non--LEP, non--collider, experiments have or
expect raw data rate of thousands of tapes per year, it is reasonable predict total CERN
data volumes of 8 to 10 times the MUSCLE single experiment estimate. This implies at least
300,000 active tape cartridges in 1992 assuming use of 200 Megabyte cartridges. In the
absence of fibre optic networking, distribution of data to universities and regional
centres will require at least a similar number of cartridges. It is reasonable to assume
that raw data are copied once at most, but master DST's will have to be copied many times
if off--site physicists are to have access to more than a fraction of their data. The
MUSCLE report set a target of 100 Gigabytes of disk space at CERN per LEP experiment by
end 1991. This represents 1.4% of the expected data volume, and would suppress a large
fraction of the tape mounts, in addition to facilitating physics analysis. It was also
urged that this disk capacity was complemented by automated cartridge handling. Such a
system would probably have a capacity of 10% to 20% of the total data volume. As argued
above, it is reasonable to assume that an optimized storage hierarchy at CERN would have
about 8 to 10 times the capacity needed for a single LEP experiment. Together with system
needs estimated at 20% of the total, this leads to a target of over 1000 Gigabytes disk
storage, and automated handling of at least 40,000 cartridges by end 1991. Automatic
migration of relatively inactive files from disk to tape--cartridge storage would be of
considerable benefit in managing experiments' DSTs. For physicists' "personal"
files, which can easily and justifiably reach hundreds of Megabytes per person, such
automatic migration is even more important. The existing CERN site network is only marginally able to support data access from
workstations, and does not support bulk data transfer between computers. Physics analysis
at CERN would benefit from improvements in both areas and efforts are underway to achieve
this. There is no physics reason why access to data from universities and regional centres
should need a lower performance than CERN--site networking. This runs contrary to
conventional networking philosophy which assumes that traffic falls off rapidly with
increasing distance. Although it is financially unreasonable to expect that wide area
networking should always offer local area network performance, the effectiveness of
off--site physics analysis will increase as this goal is approached. A recent study
[Footnote: E--A. Knabbe (CERN--DD) "What would the LEP experiments do with 2
Megabit/second link(s)?", September 1988. ] of how LEP experiments would use 2
Megabit/second links showed that the links would be used by a wide variety of services
(paralleling the wide variety of services offered on the CERN site). For many activities
remote users would have access similar to that on the CERN site, although a truly
integrated environment for the location--independent analysis of HEP data will require
much higher speeds. Although 2 Megabit/second links are not sufficient to provide totally
transparent access to arbitrarily large data samples in real time, their effectiveness can
be enhanced by having sufficiently large data storage facilities at the remote end of the
link. Frequently accessed data can then be stored at the university or regional centre.
The management of these distributed data stores could, in principle, become totally
automatic. A substantially integrated environment, supporting analysis activities
requiring access to Gigabytes of data, could be created by using networks based on optical
fibres between universities and regional centres and between the centres and CERN. The
current capacity of such links is in the range 34 to 144 Megabits/sec., and the same
physical fibres will be able to support at 4 or 16 times these rates within the next few
years. The true commercial cost of such fibre--optic networking would be low enough to
make its early use by HEP very attractive. In reality, its early use in Europe would
require a significant change in PTT tariff policies. The next five years will be a challenging period in the development of computing for
offline High Energy Physics, as we progress towards a distributed computing environment,
exploiting the fast--developing technologies of workstations and networks, at the same
time ensuring that the needs of the early years of LEP are met effectively. The main elements in the evolution of centrally organized computing, at CERN, in
national and regional centres, and in the universities, are:
It is important that this evolution take place smoothly. While we must press ahead with
the introduction of new technologies for distributed computing, we must also learn how to
exploit these effectively, and be careful not to abandon prematurely existing proven
services. There are three main sources of "production" batch capacity available to
high--energy physicists: the CERN central computing services; large national physics
institutes or computing centres; smaller systems such as departmental VAXes, emulator
farms, off--duty personal workstations, etc. By the end of 1988, the CERN Computer Centre will provide two batch services, based on
the following equipment.
These figures are the nominal capacities of the computers. In practice, a significant
and growing proportion of the IBM service will be used for interactive computing, which
consumes about 20% of the IBM--compatible capacity delivered at present. An additional 10%
of the CPU cycles are consumed by general services, such as network file servers, mail
handling, and printing. This percentage will increase significantly with the development
of distributed computing. The requirement to increase the general--purpose batch capacity
to over 150 CERN units by the end of 1992 may be met in a number of ways. This section
discusses some of the general factors involved, and suggests a model which is reasonable
today in order to enable costs to be estimated. In practice the route chosen will depend
to a large extent on future product announcements and the results of financial
negotiations with manufacturers. Competitive batch services were first introduced into the Computer Centre at the end of
1976, when an IBM system was installed alongside the existing CDC 7600. The original
motivations for installing a second manufacturer's equipment were to provide a wider
service for users, allowing them to benefit from the different strengths of the different
systems, and to provide greater possibilities for compatibility with services in outside
institutes, avoiding placing the latter in a position where they were under pressure to
buy from CERN's single supplier. These remain good reasons for continuing to operate
complementary services today. However, the past twelve years have demonstrated that this
competitive environment has additional advantageous effects: on the suppliers, in terms of
quality of service as well as more imaginative commercial terms; on the staff providing
the services; on the users, who are encouraged to avoid dependence on proprietary
technology - software, operating systems and hardware. The scale of CERN central computing
is sufficiently large for the operation of two services, and it is important for the
flexibility and vitality of the central batch services that the competitive element
remain. Evolution of the capacity implies also evolution of the operating systems which can
best provide it. There is clearly a strong incentive to provide a stable environment for
the users, and operating system changes and the introduction of incompatible hardware must
be kept to a strict minimum, while ensuring that obsolescence is avoided and a reasonable
flexibility to exploit new hardware is maintained. These conflicting requirements are hard
to reconcile. In preparation for LEP, the IBM service is being moved to a new base on the
VM/XA/SP operating system. This will ensure that the service will be able to exploit the
hardware innovations which will appear during the next few years, particularly in terms of
memory size and multi--processor configurations, at a minimal inconvenience to the user.
Similarly, the Unicos operating system on the Cray will enable either of the Cray
architectures (the X--MP/Y--MP and Cray--2/Cray--3 lines) to be used. There is therefore
no need to make major changes in operating systems on these services during the next five
years. There is, however, considerable activity on the part of most computer manufacturers and
in various standardization bodies in the area of Unix, which promises to provide a single
standard user interface on all computer systems currently of interest to offline High
Energy Physics (e.g. Cray, IBM 3090 and compatibles, DEC VAX, Gould, Apollo, IBM PC, Apple
Macintosh). While there is ample scope for argument about the attractiveness of Unix as an
operating system, we now have practical experience of good manufacturer--supported
implementations (e.g. Unicos) which demonstrate that it has no fundamental flaws (and
indeed has many advantages) in our environment. The benefits of a standard user interface
are clear, but we must carefully weigh these against other factors like additional
functionality, ease of use, and manufacturer commitment, before deciding to abandon
excellent systems such as VMS and MAC OS. IBM's AIX system for the 3090 is particularly
interesting in view of the company's declared commitment to it, and a number of
deficiencies perceived in the current versions of alternative 3090 operating systems.
However, the first release is not scheduled until 1989, and there are important features
missing, such as tape and batch support. We should actively study the AIX system when it
becomes available, and also investigate how we can best influence the various Unix
standardization activities (POSIX, X/OPEN, the Open Software Foundation). The IBM--compatible and Cray batch services are very different in many respects, the
most significant being that the Cray has a substantially higher potential vector
performance, but has a limited disk configuration (due to the cost of the
high--performance disks which alone are supported on Cray machines), and lacks virtual
memory. Today, and probably during the five--year period being considered, the vector
capability of these machines will not be of major significance in overall
throughput, although individual programs may make dramatic gains, and in turn perform
substantially better on the Cray. A brief analysis of comparative list price costs
[Footnote: The list price of a Cray Y--MP system with 8 processors, 32 Megawords of main
memory and a 128 Megaword "solid state disk" is $20M. 200 Gigabytes of disk
space cost about $10M. The total cost of a system with a capacity of about 96 CERN units
is therefore $30M, or about 500 kSF/unit ($1=1.6 SF). The list price, including standard academic discount at 20%, of an IBM 3090--600S/VF,
which would deliver about 50 CERN units, together with 256 Megabytes of main memory, 256
Megabytes of extended store and 100 Gigabytes of disk space is about 27 MSF, or about 550
kSF/unit. The list price for a four processor Amdahl 5990--1400, with 256 Megabytes of main
memory, 512 Megabytes of extended memory, and 100 Gigabytes of disk space, but no vectors,
and which would deliver about 50 CERN units is about 30 MSF. A discount would normally be
expected. ] of IBM and Cray systems which will be available at the beginning of next year
shows that these systems are rather competitively priced in terms of scalar computing
capacity per Swiss franc. Thus, while Cray--style computing does not have to be
financially justified in terms of vector usage, the inconvenience of restricted memory and
disk capacity will limit the type of work for which these machines will be suitable. It is clear that the provision of two separate batch services using different equipment
has direct manpower implications. These are most significant during the installation of
the hardware and introduction of the service. However, the style of batch service offered
on the Cray, aimed at a relatively small number of very large users, restricts the
continuing support manpower considerably, particularly in comparison with the manpower
required to run a general--purpose service like that offered on the IBM systems. About 5
staff are required to maintain the Cray service in addition to those which would be
required to run a similarly restricted service provided by duplicating the hardware and
software environment of the general--purpose IBM service. This may be considered as the
cost of running complementary batch services. Although it is not possible to quantify the
value of any single element of a financial negotiation, the competitive environment
generated by the two--manufacturer policy for batch capacity has been an important factor
in the achievement of keen commercial terms for computer hardware in recent years,
largely, if not completely, off--setting the cost of the additional manpower involved. The VM service operated on IBM and compatible computers by the CERN Computer Centre
provides a number of quite different facilities, the most important of which are: major
batch service; general--purpose interactive service; "front--end" support for
the Cray; printing and text--processing facilities for many other computers and
workstations, in particular the central VMS service; file server for workstations and
small computers. Operating this general--purpose service clearly requires much greater manpower than
that needed for a straightforward batch service, due to the variety of facilities to be
supported and the overall complexity of the system. It must also be noted that the
ancillary services consume a non--negligible fraction of the CPU capacity of the
configuration. Over the next few years, as discussed in later sections, the VM service will play an
increasingly important role as a network file server, which will require a substantial
amount of high--priority CPU capacity. COCOTIME [Footnote: The CERN committee which
allocates central computing resources to experiments.] and the Computer Centre will find
the task of scheduling resources increasingly difficult in this complex environment. IBM--compatible systems have served the industry well, both in improving the
price/performance of IBM--style computing but also in stimulating IBM to build powerful
processors which perform well in the scientific and engineering markets. However, the
competitive response of IBM has also included the provision of obscurely specified
hardware and microcode enhancements which are essential for the efficient operation of the
machine using IBM software. For example, the new VM/XA/SP operating system cannot operate
on CERN's Siemens 7890S computer, installed less than four years ago. Great care must
therefore be taken in assessing the overall financial benefits of IBM--compatible systems.
The DEC services in the CERN Computer Centre have been set up in order to provide
interactive computing facilities. However, due to the fact that an interactive service
must cater for peak daytime loads, some 60% of the installed capacity is available for
batch. At present this amounts to only about 4 units, including the capacity of the
engineering service VAXes, but the recommendation below for the expansion of the
interactive service could lead to somewhat greater capacity being available for batch.
This will be of interest to small experiments which are thereby able to do all of their
computing on VAX. The capacity available will not, however, exceed 10% of the total
general--purpose batch capacity at CERN, and therefore does not justify the operation of a
full--scale batch service. DEC is not considered here as an alternative to IBM--compatible and Cray for the main
central batch facilities, because of the large number of CPUs required to achieve a
reasonable capacity, although the cost of an appropriately configured system is comparable
with the others. [Footnote: A cluster of four of DEC's largest systems, the
four--processor VAX 8840 with 128 Megabytes of memory, together with 50 Gigabytes of disk
space and 4 disk controllers, costs about 10.5 MSF with standard discount. This should
deliver a nominal total of 24 units on sixteen processors. The
multi--processing efficiency of the computers and our ability to schedule 16 processors
efficiently for general--purpose batch are unknowns, but we might expect a degradation in
throughput perhaps as high as 20%. The cost per CERN processing unit is therefore in the
range of 440k SF to 550 kSF.] While we can expect more powerful CPUs to be delivered by
DEC over the next few years, the other manufacturers will also introduce new equipment,
and it is not clear that the relative position will change significantly in the short
term. Replacement of the Siemens 7890S The first step is the replacement of the aging and
obsolescent Siemens 7890S, which is not capable of running the latest version of the VM
operating system, and which will provide less than 25% of the IBM--compatible capacity
after the installation of the IBM 3090--600E system at the end of 1988. The Siemens should
be replaced at the end of 1989, about six months after the startup of the LEP machine, by
a system with at least 50 units of scalar capacity [Footnote: This will bring the total
IBM--compatible capacity to 90 units, of which we assume 20% will be used for
general--purpose interactive work.] This could be purchased from IBM itself, or from a
compatible supplier. Clearly, the chosen system must run the latest version of the VM
operating system, with reasonable confidence that it will not become obsolete in this
sense during its four--year lifetime. Possible candidates today would include a four
processor Amdahl 5990--1400 or an IBM 3090--600S. Assuming that a 40% discount over list price could be achieved, the cost of this
operation will be around 18 MSF. Note that the existing IBM computer, installed originally early in 1986, will itself
have to be replaced by the end of the period being considered. The Cray system was installed in the spring of 1988, and it is therefore
too soon to assess fully the effectiveness of this style of service for the different
types of processing required for LEP physics. It is clear that it will be better suited to
work which is CPU--intensive, and that a very large disk capacity is excluded due to the
high cost of the disk equipment. The I/O limitations are, however, sometimes
over--emphasized - the system has excellent tape support and the 50 Gigabytes of disk
space which should be available early next year will enable a good tape staging service to
be provided to support most Monte Carlo and production programs. Vector support will become available on almost all high--end processors over the next
few years for relatively little additional cost. The ability of high--energy physics to
benefit from such hardware would bring significant financial savings, even if the overall
improvement in throughput is relatively low, due to the high total cost of the investment
in such a large system. In order to stimulate the pioneers in vectorization a machine with
an outstanding vector performance should be maintained beyond the life of the present
Cray. Subject to the satisfactory development of the Cray service, the current X--MP/48
should be replaced during 1991 by a system with at least 80 units of scalar capacity. The
desire to minimize change will make the Y--MP system an obvious candidate (the eight
processor model should deliver over 90 units). However, it may well be more attractive to
install the Cray 3 system, if it is available at that time, providing a substantially
better scalar performance with its GaAs technology, and a very large real memory. Clearly
any other comparable supercomputers which may become available must be carefully
considered, with due attention paid to the importance of the needs for a competitive
environment and minimal change of operating system mentioned above. The system should
include at least 100 Gigabytes of disk space, and we estimate a total cost of around 30
MSF, assuming a substantial discount. The evolution of the disk capacity of the IBM--compatible service is
covered in a later section on data storage. CERN Computing Centre - Evolution of Batch Capacity The nominal CPU capacity of the IBM service has been reduced by 30% in 1988, rising to
40% in 1991, to cover the part used by general--purpose interactive and distributed
computing services. In order to make a major impact on LEP data analysis, regional computer facilities must
include very large disk configurations. In turn, to make efficient use of this expensive
investment and justify the personnel costs which will be necessary to manage the DST data
and associated calibration files, substantial CPU capacity must also be available. Because
of the practical management difficulties involved, an individual collaboration will be
able to organize its large--scale computing at only a limited number of physical sites. We
conclude that a centre which will make a significant contribution to several experiments
must provide minimum resources of 100 Gigabytes of disk space, an automatic cartridge tape
loader with a capacity of 5K tapes, and a CPU capacity of 20 units. [Footnote: This
represents an investment of about 13.5 MSF (100 Gigabytes cost 3 MSF; a one third share in
a tape loader costs 0.5 MSF; 20 units of IBM CPU power costs about 10 MSF at academic list
price). ] Increased CPU capacity would make increasingly effective use of the disk space,
and, if adequate network connections to CERN are also available, [Footnote: See a later
section for discussion.] the centre could be a very attractive focal point for a
substantial part of the analysis of an individual experiment. We therefore suggest that the conventional batch computing capacity available to high
energy physicists be concentrated in a number of large, adequately configured, regional or
national centres, to provide a total processing capacity of at least 200 units. The experience with the 3081E and 370E emulators, and at Fermilab with the ACP, has
already demonstrated that these systems can be used effectively in many cases where the
program is relatively stable. In order to maximize the cost benefit of such systems, it is
important that they are provided with an adequate level of operational support, including
tape handling. At the low--end of the CPU market, the cost of raw processor power is falling at an
astonishing rate, which will continue to stimulate the development of cheap parallel
processor systems. These range from straightforward assemblies of micro--computers which
could be exploited in the same way as the existing emulator farms, to very specialized
systems such as the Meiko Computing Surface. A number of parallel processor "farms", established in universities, regional
centres or in the CERN Computer Centre, well integrated with the institute's central
computing services, to provide easy and cost--effective access to the main data storage
facilities, could make a major contribution to the general computing capacity of HEP. Once
established, these services would provide a framework which could later be very rapidly
expanded in capacity, for relatively small additional cost. We therefore make the following recommendations.
We note that substantial batch processing capacity will exist in the form of private
computing facilities, owned and operated by individual collaborations. Much of this will
be in the form of emulators and other parallel processors, often purchased for a specific
task associated with data collection but capable of being used for more general offline
analysis at other times. These facilities are outside the scope of this section of the
report. The aggregate capacity of the small computers installed in university physics
departments will grow substantially over the next few years. Some fraction of this will be
available for data analysis, depending on the ability of the physics departments to supply
data to the computers and of the collaborations to organize the distribution of work to
the many small units. The main contribution of these is likely to be as tools for a
department's own staff, thereby enabling them to perform more of their work while at home.
See also the discussion in the later section on interactive data analysis. The number of personal workstations installed at CERN will grow dramatically over the
next few years, and may reach a total of 1000 within four years, representing a nominal
CPU capacity of several hundred CERN units. Since the workstations will be under the
control of the individual users it is unreasonable to imagine that any sensible fraction
of this capacity could be used by a central service. However, individual experiments
should be encouraged to take advantage of spare capacity on their workstations for
suitable tasks such as Monte Carlo. General--purpose interactive facilities include a program--development environment,
text--processing and document handling facilities, electronic mail and database services,
together with good connections to the main local and wide--area networks, to ensure that
the user is able to control effectively his activities on central services at CERN and in
regional centres. To an increasing extent, interactive services and facilities will be
used for data analysis which, because of the rather special requirements for CPU power,
data storage and network bandwidth, is discussed separately in the next section. The following discussion refers specifically to CERN, but it applies also, with
appropriate scaling, to regional centres. The benefit to physics analysis of a rapidly expanding use of personal workstations has
already been discussed. In order to ensure that the evolution towards a personal
workstation for every physicist and engineer at CERN may proceed as efficiently as
possible, we make the following recommendations.
While it is clear that there will be a steady and rapid growth in the use of personal
workstations, the life of the central time--sharing service is far from over. As we have
noted above, workstations are today cost--effective only for intensive users. The demand
for the central VM and VMS services continues to grow, and shows no sign of diminishing.
We expect that the move of users from the central services to workstations over the next
two years will be more than balanced by the increase in the number of new users of the
central services, and that the demand at peak times will double during the same period,
which will see the growth in the management information and engineering services, and the
startup of LEP. Thereafter, we expect to see the rate of increase diminish, and demand
eventually tail off. The central interactive services are at present provided by the VM service (at the
level of about 20% of the installed IBM--compatible capacity), and by the VMS service
(which has been installed specifically for this purpose). The popularity of both of these
services indicates that they are effective and complementary, and this and the advantages
of a competitive environment enumerated above lead us to recommend that they should both
be maintained and developed. The current load problems of the VM service will be largely
resolved with the introduction of the VM/XA/SP operating system which will enable the
service to expand to take advantage of additional processors and memory which are already
installed. It is assumed that the VM interactive service will continue to use up to 20% of
the installed IBM--compatible capacity, and that no specific recommendation for expansion
is required. The VMS service is, however, badly overloaded at present, and an effective increase of
around 60% in its capacity is required merely in order to provide an acceptable service.
This should be installed without delay. In addition, the capacity of the service should be
doubled before the end of 1989. The total cost of the initial upgrade, including a modest
10 Gigabytes of disk space will be 1.4 MSF. The cost of the subsequent doubling in
capacity, replacement of obsolete equipment and growth of the disk capacity to a total of
100 Gigabytes, is likely to cost about 7 MSF at standard prices, over two years. The growing use of personal workstations implies a growing demand for centralized file
and CPU server facilities (increased productivity will lead to increased consumption of
resources as well as increased output). The VM and VMS central time--sharing services are
the natural vehicles for such support, and as users move towards workstations they will in
any case go through a phase where their working environment is split between the central
service and the workstation. We can be sure that any new investments in the central VM and
VMS services will be of long term value for workstation support, even if the time--sharing
demand drops faster than we predict. The use of more or less powerful workstations for interactive data analysis gives rise
to the following requirements, which are in addition to the general needs of workstations,
discussed in the previous section. These requirements are stated in terms of CERN, but it
is clear that they can be extrapolated to apply to the larger national and regional
centres used by HEP.
All current models for LEP data analysis assume that the master
"data summary tape" (DST) will be stored on cartridge tapes managed by the
central VM service, with successive levels of subsets (team and personal DSTs) stored on
disks on the same system. In some cases it may be possible to pre--select a relatively
small subset of the personal DST (say 5--10 Megabytes) and transfer that to the
workstation for processing during a working session of a few hours. This mode of working
can be accommodated within the relatively modest data rates required for basic workstation
support. However, in cases where this pre--selection is not practical all of the external
data consumed by the analysis program must come from the IBM file server. This gives rise
to a requirement for very much greater data rates than can be achieved by any existing
product. The minimum requirement is binary file access at an aggregate data rate of 500
kilobytes/second per cluster of 10 workstations of the VAXstation 2000 or Apollo DN3000
class, rising to 1 Megabyte/second by the end of 1989. The demand will continue to rise
faster than the technology available. The processing power of the workstations will
continue to grow at a rate far outstripping developments in deliverable network bandwidth.
It will be necessary to envisage the installation of the highest performance network
facilities available, specifically for the support of the data analysis workstations. It
is important that the development of distributed physics computing is not restricted by
the considerations of a general--purpose network infrastructure. It is likely that the
coming 100 Megabit/second FDDI (Fibre Distributed Data Interface) standard will be an
appropriate choice in the medium term. In order to benefit from the different characteristics of
central mainframes and workstations, and to minimize the effects of practical constraints
such as insufficient network bandwidth, considerable investment should be made in the
development of co--operative processing techniques. Already the Physics Analysis
Workstation (PAW) [Footnote: PAW - Physics Analysis Workstation, Comp.Phys.Comm. 45, 1987;
The PAW Users' Guide, CERN Program Library Writeup Q121.] developments permit the
separation of the user interface (on a workstation) and the Zebra data server (on a
mainframe). Future plans envisage the extension of these techniques to provide
"object servers" which may be distributed with some freedom between the
workstation and specialized central services such as the Cray (for CPU--intensive tasks)
or the IBM (for data storage manipulation). The effects of this style of working on the
scheduling of the central resources may be profound, but the problems raised must be
tackled with vigour both by COCOTIME and by the technical staff involved. Co--operative
processing will become a major factor in the analysis of LEP data. We have seen that the centralization of DST data poses a
major problem for distributed computing. It may be necessary to consider the use of
distributed file servers, in order to relieve the bottleneck of the central IBM file
server. We must note, however, that the task of handling the DST data for LEP requires
considerable CPU capacity [Footnote: The MUSCLE Report estimated that an IBM 3090--600E
will be required to perform this task for the four LEP experiments in 1991.] , and so a
distributed file server will be itself a substantial system, resembling more a mainframe
than a workstation. The decision to install distributed file servers is largely a
financial one, and at present the economics favour IBM--compatible systems, particularly
due to the price of disk storage. In the immediate future, therefore, we recommend that
the task of providing a central file server facility should be the responsibility of the
IBM/VM service. It is important that collaborations include estimates of their needs for such
facilities in their computing requests, in order to ensure that appropriate funding and
planning may be organized. Note that the figure of 20% of the IBM service which was
allocated above to general--purpose distributed computing support does not, by a
very substantial amount, include the costs of support for interactive data analysis. It is
assumed that this is included in the quoted batch CPU requirement of 160 units. There are at present about 100 Apollo and 100 VAX workstations
installed at CERN. We have been experiencing exponential growth, but it is expected that
the rate of increase will settle down to about 200 new systems per year, leading to a
ceiling of some 1000 stations. The useful life--time of a station might be expected
to rise from about three years today to five years by 1992. We expect about 10% of the
stations to be "high--end" systems costing an average of 80 kSF each, while the
others will cost on average 30 kSF each, giving a total annual investment of 4.3 MSF.
Central support costs, including special servers, gateways, software licences, etc. will
bring the total annual cost to around 5 MSF. Maintenance charges for these systems are at
present very high, amounting to about 25% of the (heavily discounted) purchase price per
annum. The most efficient way to staff workstation support activities will be to provide a
well--managed and strongly coordinated central service, rather than using large numbers of
part--time people with a variety of employers, interests and priorities. Estimates for
central staffing are as follows, assuming that half of the workstations are Unix--based,
and the other half use VMS. As the use of personal workstations
for interactive data analysis evolves, they will become a requirement at all institutes
participating in the analysis. Large centres with substantial disk capacity will therefore
require to provide services to support these workstations and integrate them with their
central processing and file server facilities. The larger institutes and CERN should also
provide software and systems management support services for the smaller institutes. Once distributed co--operative processing is established as the normal way of computing
for offline physics at CERN and in the major regional centres, there will be a strong
demand for improved integration between these centres and CERN, and with smaller
institutes and individual physics departments. We should be aiming for the development of
distributed data analysis on a European--wide basis, with the individual physicist able to
work effectively from his personal workstation, using databases and DST files located at a
number of different institutes. This will require not only a high level of coordination
between the main computing centres, but also high--performance reliable networks. The
requirements for the latter are covered in the following section. The distributed computing environment which we envisage will
highlight the inefficiencies of using different floating point number representations on
the various computers involved. Although most workstations now use the IEEE standard
format (with the exception of VAXstations), each of the mainframe types commonly used
within high energy physics uses its private format. This report 'Computing at CERN in the 1990s' includes a part dedicated to general
questions of networking. This section is therefore restricted to a brief summary of the
requirements of centrally organized computing. The on--site Ethernet Service must continue to grow and be improved in a number of
areas.
The Domain token ring infrastructure for Apollo workstations must be maintained, with
appropriate bridges to the Ethernet infrastructure and central services. The medium term (the next three years) will see the migration towards 100
Megabit/second FDDI token ring, initially as a "backbone" interconnecting the
Ethernet and Domain infrastructures and the central services. An early priority in this
development, which may well dictate the topology of the initial network, must be to ensure
high bandwidth connectivity of workstations used for physics data analysis and the central
IBM/VM service. It should be assumed that within three years, most new physics analysis
workstations will be of sufficient power to require connection directly to the FDDI
network. [Footnote: By that time, the cost of an FDDI connection should be comparable with
that of an Ethernet connection, making it the network of choice if the cable is
available.] We should also expect to see FDDI coming into use to connect remote peripherals to the
central mainframes (e.g. magnetic tape units installed in experimental areas), and as a
means of interconnecting mainframes from different manufacturers within the Computer
Centre. In preparation for the longer term, CERN should acquire experience with the 100
Megabytes/second High Speed Channel (HSC), [Footnote: ANSI Committee X3 T9.3.] which will
initially become available as a means of connecting very high performance graphics
workstations to supercomputers. It should also, however, develop as a standard
inter--processor channel suitable for the Computer Centre. The EARN and HEP-- DECnet networks which have developed over the past few years with
emphasis on connectivity and guaranteed delivery provide good general--purpose facilities
suitable for remote job entry and mail, and for the transfer of small to medium files. The
European--USA links provided are very important features of these services. Both DECnet
and the HEP X.25 network also allow elementary remote login. The ability of EARN to
increase its services to include full--screen remote login and inter--process
communication facilities will be major factors in its continued success, as will the
improvement of the capacity planning for its backbone network. In the event that a
pan--European network for research and development emerges from the Eureka COSINE project,
an adiabatic transition from the above services, including transatlantic gateways, would
be vital. Effective high--bandwidth connections between CERN and the major national and regional
centres are required by the end of 1989, in order to enable the efficient management of
physics production and data analysis work at these centres and at CERN. These links will
provide database access and file transfer (of calibration data and programs, and perhaps
small samples of physics data), in addition to remote job entry and other facilities.
Since much of the communication would be automated, such as direct access from programs to
controlling databases at CERN, reliability is vital. Efficiently operated 2 Megabit/second
links will be required at least between CERN and Aachen, RAL, IN2P3, Saclay, and an
Italian site (Bologna?) and in due course also CIEMAT, DESY, NIKHEF and the USA.
Experience with such links should be obtained as soon as possible. It is likely that
neither HEP, nor the wider European research community would benefit from a completely
shared network for all activities. The high traffic rate needed by HEP for distributed
data analysis is inappropriate for general purpose networks. However, low--level
multiplexing of HEP and general purpose traffic over the same fibres does seem a promising
approach. The use of personal workstations will extend the user's normal computing environment to
his office desk, in a form which he will be able to duplicate conveniently in his home
institute and at CERN. The degree of integration with centralized file and CPU servers,
which the CERN or regional computing centre offers him locally via LANs, will become an
increasingly important factor in his productivity, giving rise to a demand for
general--purpose communication links between CERN and the major computing centres in the
member states, supporting 10--100 Megabit/second data rates. The current proposed
experiment to connect CERN to several German laboratories using 10 Megabit/second channels
on a 140 Megabit/second PTT fibre will give useful experience in this area and should be
actively supported by CERN. Personal workstations, and the development of distributed computing which they will
stimulate, offer the first real opportunity for physicists to work effectively away from
the CERN Computer Centre; the real cost of high--performance networking is now
sufficiently low to enable them to do this from their office at CERN, or from their desk
in their home institute; HEP must ensure that the tariff policy of the European PTT
monopolies does not prevent European physicists from taking advantage of this situation. This section summarizes the requirements described earlier in this report.
The DST data required for LEP physics analysis, as estimated in the MUSCLE Report,
poses two major problems for storage and distribution.
Thus, managing the LEP DSTs will require a lot of disk space, serviced by a
high--capacity cartridge tape robot, and supported by powerful CPUs. Economic (the cost of
disk space and CPU power, and the possibility of sharing robots and tape drives) and
practical (ease of management and operation) arguments will favour a centralized solution,
at least for the first phase of LEP. The principal drawback to such a solution is the
difficulty in achieving sufficient network bandwidth to feed the data to the workstations
where the final processing will take place. This is mainly due to the current relatively
poor throughput of heterogeneous network connections to IBM, rather than to any
fundamental weakness in the concept of a centralized file server. [Footnote: DEC LAVCs
achieve sustained aggregate network data rates of around 250 kilobytes/second, more than
twice that available with a single Ethernet connection to the IBM. Even higher rates can
be achieved between some 80386 based PC--s. This is due to the use of light--weight
protocols and efficient implementations of them. While the former may not be easily
applicable in a heterogeneous situation (where standard protocols must be used), we can
expect to see the development of more efficient implementations, and the use of
"protocol engines" in the hardware of the communications interfaces, in a 2--4
year timescale. In the meantime, it seems reasonable to connect multiple Ethernets to a
central file server, using several physical controllers, in order to achieve higher
aggregate throughput.] We must study carefully early experience with LEP data analysis using workstations, in
order to understand better the issues of data distribution. In the meantime, strategies
for file servers should remain flexible and be reviewed regularly in order to be able to
benefit from changes in storage costs and technology. During the next five years, improved wide--area networking will enable individual
events, or even occasional complete tapes, to be transferred between CERN and external
institutes with some convenience, but until optical fibre bandwidths become available,
almost all of the DST data will be distributed on cartridge tapes. In order to assist in
this massive operation the following central services should be provided.
The cost of tapes and freight should be charged to the requestor of the data, probably
through a CERN team account. This report has already described the need for a data management for the whole CERN
community. This problem is too complex, and depends too greatly on the situation at
regional centres and universities, for a technical solution to be proposed here. Technical
solutions, almost certainly starting as small scale trials rather than global services,
should be pursued by a committee reporting to the management of European HEP. A more
specific recommendations can be made about a more limited goal. The need for automatic
disk--to--tape migration of user files and experiments' DSTs has been described in section
In this chapter we address the questions of software support. We concentrate on
system--related software and on general questions of software support. The following
chapters will go into more detail on specific areas where software support of an
application--related nature is seen as an existing or potential problem. The distributed computing environment described in this report could revolutionize
physics analysis. However, the increasing complexity of the hardware infrastructure must
not be allowed to produce increasing complexity for the user. Every effort must be made to
render the appearance of the operating system, and of the data storage, as
location--independent as possible. At the same time as hardware complexity is increasing,
so is the complexity of HEP software. The software for any current large experiment is
well beyond the point at which any single person could understand all the details. A later
chapter will describe the promise of software engineering techniques for the design of HEP
software. In addition, it seems inevitable that coding will eventually have to be done at
a far higher level of abstraction than that of current 'high--level' languages such as
Fortran. The creation of an HEP--specific 'even--higher--level' language is an option
meriting further study. An environment in which physicists must use only one viable, but unexiting, operating
system may well be more productive than one in which they must use several better, but
different, operating systems. This is the context within which HEP should give serious
consideration to an increased use of Unix. For all workstations except those from DEC,
Unix is now the default operating system, and on many workstations there is no other
possibility. At CERN the CRAY runs under Unix, and VAXes and VAXstations could (but
usually don't) run under Unix. However, until now, Unix has not been available for the IBM
systems which are the centre of data--handling and computing for most experiments. With
the announcement that AIX, (IBM--Unix) will become available for IBM mainframes during
1989, it becomes possible for physicists to contemplate using only one operating system.
The "Open Systems Foundation" initiative, involving Apollo, DEC and IBM among
others, offers the hope that the different implementations of Unix will really appear very
similar to users. It would certainly be premature to recommend that HEP convert rapidly
and totally to Unix. However, it would be unrealistic to recommend that no Unix service
should be offered on central IBM systems until its success was assured. Few physicists
have any experience of being able to do all the analysis of a large experiment within one
operating system, and an abstract study of its advantages would almost certainly be judged
inadequate. It is recommended that an AIX service be started on the CERN IBM system as
soon as the AIX specification becomes adequate to meet HEP needs. This service should be
started on a small scale and its expansion should be driven by user demand. Some
flexibility will be necessary in assessing whether AIX has reached an adequate state of
development. It is not recommended that CERN write batch and tape handling systems to
provide essential facilities missing from early versions of AIX, but smaller deficiencies
might be repaired by CERN effort. This recommendation implies that some new manpower must
be found, or that there will be some degradation in the support for the existing CERN
services. It should also be noted that, even though great benefits are expected from the
use of a single operating system, exchange of data between systems will still be hindered
by their different data representations, except between systems using the IEEE/ASCII form The working group has not attempted an evaluation of alternatives to Fortran, or even
of the impact of the likely evolution of Fortran. For all its inelegance, and lack of
safety features, it seems certain that Fortran will remain the main language for HEP code
well into the 1990s. No existing or planned language, including, of course, Fortran,
provides machine independent tape or network I/O. HEP has overcome this problem with
utilities such as the ZEBRA--FZ package, which are effective, although naturally less
convenient than if they were integral parts of the language. Fortran is notoriously
lacking in data--structure concepts. Many more modern languages have some support for
data--structures, although it is not clear that they could support the complexity required
for handling HEP data. Fortran at least allows the freedom to provide sophisticated
data--structure support through additional packages such as ZEBRA--MZ. The ensemble of HEP
packages which run on top of Fortran, together with Fortran itself, form a powerful
environment for HEP data analysis. It would not be easy to guarantee that a move to
another language would improve the overall environment. Most physicists do not have the
time or background knowledge to perform a thorough review of languages before starting an
experiment. The CERN or HEP community should periodically review developments in
languages, both to add weight to their input to Fortran standardization committees, and to
identify any serious alternatives to Fortran. Ten years ago, most experiments at CERN expected and received support from CERN, or a
major laboratory, to write the experiment--specific software. This help normally took the
form of programmers assigned to the experiment, and was especially important for
real--time programming and data--acquisition. Today, CERN and the major laboratories are
not the sole repositories of computing expertise, and the case for this type of software
support has weakened. However, there remain some good arguments for the provision of some
experiment--specific support. Even in the largest experiments continuity is a major
problem; the efficiency of the software effort can be greatly enhanced if at least some
people can work full--time for many years. It is very difficult for universities to
provide these people, especially if they must work largely at CERN. In the case of smaller
experiments with shorter timescales, it is inefficient to require that, for example,
university physicists become experts in a standard data--acquisition system which they
will only use for a few months. The current level of experiment--specific support is
widely criticised by physicists. This criticism may reflect the unclear (or non--existant)
algorithms used to allocate support personnel, and the lack of coordination between CERN
and other centres, rather than truly inadequate resources. CERN, and the regional centres
and laboratories which can also provide such support, should review urgently how support
is organized, and how this organization should be improved. Central support for widely usable packages and libraries has been extremely successful
for many years. The CERN libraries, and packages like GEANT3, HBOOK and ZEBRA are used not
only within the CERN community, but also for many HEP experiments worldwide. A relatively
small centralized effort has prevented much wasteful duplication, and has made high
quality software available to experiments which could not afford to write it themselves.
It is very unlikely that a similar success could have been achieved merely by hoping for
cooperation between experiments. In spite of this evident success, there are still some
problems:
An additional, and even more complex problem, is the determination of which software
should be bought commercially. There can be no doubt that widely available commercial
software, meeting the needs of HEP, should not be duplicated. The success of less mature
commercial products, where HEP is, perhaps, a major fraction of the market, is much less
clear, and existing experience does not make it easy to formulate a simple policy about
what should be bought and what should be written. These decisions should be made on a
case--by--case basis by a technically competent body representing the CERN community. The
existing coordination problems, particularly in relation to CERN and major centres, make
it impossible to say whether manpower levels are grossly insufficient or not. It is
recommended that a software support committee be set up by European HEP with the brief to:
The committee should not limit its consideration to simple extrapolations from existing
packages and libraries. The distributed computing environment, and the increasing
complexity of HEP software make it necessary to explore new areas for HEP software
support. Examples of such areas are support for a location and device independent
user--interface to data, and possible HEP--specific very high level languages which could
simplify the coding of data processing and physics analysis. The questions specific to data base systems have been looked at by a separate group,
with the following guidelines:
Our group has used as means of contact with different groups a widely circulated
questionnaire, followed by detailed contacts with selected users, and all our conclusions
are based on this input. L.Adiels, F.Bruyant, G.Gopal, A.Osborne, P.Palazzi, E.Rimmer, and
M.Sendall have contributed substantially. We have observed a large amount of duplication of work in this area. We therefore
present clear recommendations in an attempt to limit the development of data base
applications to a small number of acceptable solutions. We are conscious of the fact that
the present recommendations still need further discussions with both users and the
designers of data bases, before final solutions can be adopted. Whenever specific products (e.g. Oracle or ADAMO) are mentioned in the recommendations
this is not to be taken literally. They have been chosen since they are used at present
and define the minimum requirement we have for any future product. We are fully aware of
the rapid development in this field. In this report, the word "Data base" is used in a broad sense and matches
with the definition given in the ECFA report: [Footnote: Databases and bookkeeping for HEP
experiments, ECFA Working Group 11, ECFA/83/78 (Sept 83).] " "A collection of
stored operational data used by the application system of some particular
enterprise." " In principle a data base need be no more than a set of computer--based records, but the
term is generally used to signify that the data are stored in separate files various types
of record different purposes is possible. The purpose of this information is to keep the name, address, telephone number both at
CERN and in the institute for administrative purposes. The integration of this information
with electronic mail addresses has been made by a few experiments. Among the solutions adopted, some experiments have no mailing list, others use card
image files with command files to use them, later experiments like Aleph and Delphi use
the NADIR system. NADIR makes use of the Oracle relational data base system to store
experiment members names, addresses, telephone numbers and electronic mail addresses.
Interactive updates and queries are performed via Oracle full--screen forms. Various ADAMO
tools are used to extract the Oracle information into portable files which are then
exported to member institutes which do not have the Oracle system. ADAMO--based tools are
then used to reconstruct and access local direct--access files and include an electronic
mail distribution system called AMSEND. One experiment uses HYPERCARD on a MacIntosh. Most experiments express their satisfaction with the adopted solution, even card image
files when the number of participants is sufficiently small. It should be noted, however,
that these data are purely for internal use. A single data base collecting all CERN users
could be beneficial to all. But the latter question was not raised in the questionnaire
and is thus the interpretation of our group only. The aim is to allow to send mail electronically to a user's family name, or to lists of
people, without the need to know his account(s). This mail is automatically converted to
his preferred mail address and optimally routed. This tool is heavily used to inform the
members of a collaboration rapidly and easily. The presently most frequently used method to retrieve electronic mail addresses seems
to be NAMES under VM. Two LEP experiments (Aleph and Delphi) are using information stored
via NADIR and used in AMSEND, developed in Aleph. NADIR stores, as well as geographical
addresses, the collaboration members home computer, his login id(s), networks and mail
servers. Other experiments rely on simple files. All experiments found their system
adequate and preferred that the updating of both the mailing lists and the electronic mail
addresses be made directly by the experiments (e.g. by the group secretaries). The DD division has set up an Oracle based system called EMDIR (electronic mail
directory) where each user is invited to enter his preferred electronic mail address. This
system is unpopular (it has a poor user interface), many users are not keeping it
up--to--date and because of this it cannot be relied upon by collaborations. None of the
experiments which replied mentioned that they are using it, although the authors report
"thousands of queries per month". They intend this to be "the basis of
future mail systems". In summary, as is the case for the geographical mailing lists, the information is used
exclusively within the experiment itself. The DD division has recently implemented an Oracle data base of the login identifiers
on the central systems of all registered central computer users and this data base
includes the CERN telephone directory. This perhaps could become one part of a common
geographical and electronic mail addressing system for all CERN experiments where the
experiments themselves would furnish the other information about their collaboration
members. The purpose of bookkeeping is to keep track of data taking runs, experimental data,
storage media (disks, tapes) and program versions used in the different phases of event
reconstruction and analysis. Amongst the current experiments, several use simple systems, like card images, others
use random access files with KAPACK or the Oracle RDBMS and some experiments have just
started with Oracle or will start soon and are still undecided. Opal intends to use Oracle
to hold "log--book" information. It seems that this is a place where a common
effort could lead to a general solution. There is no single set of data base tools today to span the whole range of on--line
applications such as storing detector constants and calibration data, histogramming, error
logging and run parameter logging. This has led to a large variety of individual solutions
ranging from simple files to KAPACK based applications or even the usage of a RDBMS such
as Oracle. The CP--LEAR experiment will use Oracle online on a VAX but only as an indexing
system pointing to other logical and physical entities (histograms, tables and files). The aim is to provide a detailed geometrical description of the detector components and
their evolution in time. It is needed in the off--line programs for the simulation and the
reconstruction of the events and for graphics display. The system must be able to cope
with a simple description, as used in reconstruction, and a very detailed one, as used in
the simulation. A fast access time is crucial, given that the simulation or reconstruction
will frequently access it to get answers of the type "where am I" and "how
far can I go". Again, a wealth of homemade solutions have been developed. Some experiments use random
access files based on ZBOOK, KAPACK or ZEBRA. The manpower investment in this area is very
large (several man--years per experiment, on average). The main difficulty resides in the
extremely fast access required. This explains why different experiments have tried to
optimize the data storage to their needs by using different solutions. It also excludes
the direct usage of Oracle for the data access at execution time. The need for a DBMS to provide fast access to Detector Description data was identified
very early on in Delphi. At first (1984) they developed a system based on KAPACK, tailored
specifically to this particular application. So long as the originally designed
Data_Structure was adhered to the system worked very well. However, Delphi were not happy
to freeze the design of the Detector Description Data_Structure so far ahead of
data--taking. So, a generalized DataBase Management System - CARGO, also based on KAPACK, was
developed, care being taken not to tailor it to the needs of any specific application. The
Delphi Detector Description - Geometry Data at first (1986), Calibration Data later (1988)
- has been successfully structured according to the conventions of CARGO. Fast access to
this data for Simulation, Analysis and Graphics programs is provided via an application
specific access package (DDAP) which structures the data requested by the User Program
into a ZEBRA division. The Geometry File contains the detailed Geometry for each of the 18 or so
sub--detectors including the beam pipe, coil and the cable duct areas. A first version of the calibration constants have now been installed in the calibration
file for almost all the sub--detectors and programs are now beginning to use this data. The L3 experiment has built a detector description data base using the ZEBRA (CERN data
structure management package) direct--access package RZ. The RZ package allows a stored
ZEBRA bank to be identified by up to 100 4 character long keys. L3 has written a Fortran
data base package (called DBL3) on top of Zebra RZ to provide the user interface to their
RZ data base. Their system stores changes to data as update records rather than changing
base records and they also store pointers (in the form of file names) to their multiple RZ
files. The user of DBL3 is normally returned a ZEBRA pointer to his requested data, the
system having read it from disk into ZEBRA memory. DBL3 took about one man--year to build
and they feel it will provide the functionality they need for many years. In addition they
have built a package for interactive access to the data base using the KUIP (Kit for an
User Interface) component of the PAW (Physics Analysis Workstation) system. Since PAW is
itself based on ZEBRA and RZ this was a natural extension. The Opal experiment will also use an RZ based system for detector description but make
the general remark that parts of their data would be suitable for a relational data base. A general solution for experiments needing the functionality of a commercial data base
such as Oracle but having constraints of performance or portability may consist in a
hybrid solution of maintaining the master information in Oracle and providing translate
programs that generate normal Fortran files. This is the solution adopted by the Aleph experiment who store their detector
description constants in an Oracle data base including validity time--stamps. The data
definition is made using data modelling with tools from the ADAMO package. The data are
then extracted into RZ or BOS (the DESY data structure management system) banks which
themselves can be stored in Fortran direct--access files. Each Oracle table becomes a
BOS/RZ bank with the numbers of rows and columns stored as the first words of the bank
(this amounts to extracting a single "view" from the relational data base). The
Oracle data base contains data from the different detector components (including the beam
pipe, cables etc.) and updating is done by 1--2 people for each of them. The purpose of this package is to store the calibration constants and to keep track of
their changes as function of time. Similar solutions are adopted as for the detector geometry, since the two need to be
linked together, the main difference being that the volume of calibration data is normally
much larger than that for detector description. Many of the same comments apply as for the
previous section. Opal will use RZ direct--access files, while Aleph will include in their
Oracle data base time stamps and pointers to their calibration data (e.g. file names or
tape numbers). All the answers received so far were related to the management of the event data in
memory. All existing memory management packages are mentioned (BOS, HYDRA, ZBOOK, ZEBRA),
though most of the new experiments tend to standardize on ZEBRA. None of these
applications involve a real data base. In particular no experiment mentioned the need of a
data base for direct access of event data. Opal envisages an event server which can recall
events from specific blocks on 3480 cartridge tapes and expressed a wish for a tape
management system at the computer centre. The intention was to find out whether people feel the need to keep track of program
versions in time, in order to allow the reprocessing of events in the same conditions (and
with the same bugs) as they were a long time ago. Also here, the question was not clear
enough and all answers were related to source code management. The purpose is to keep a catalogue of results/presentations for on-- and off--line. In
the on--line context, this may be monitoring histograms, and in the off--line context,
reconstruction results or physics data. In the answers, a majority expresses support for PAW. A data base could keep the whole software documentation and the description of the
data. Several experiments use simple files. The only applications using data bases are from
Aleph using the data model tools from ADAMO to provide software documentation (in addition
to write--ups on paper). Opal uses an internal system called OpalDOC but would merge this with the DD CERNDOC
system when some bugs in it had been fixed and when it accepts TEX formatted files. The purpose is to have an up--to--date list of all publications for the experiment.
This should include the title, the list of authors, the reference number, the data and
subject keywords. Currently, the experiments use simple homemade solutions. They are satisfied with their
solutions, but some wonder whether in the long term a more elaborate system may not be
preferable, provided the access remain sufficiently simple. The Particle Data Group uses
Oracle. The EP Electronics are using ISIS and find it hardly adequate. A need expressed by experiments and mechanics groups is to have a data base application
which can handle the registration of parts of the equipment and their life history. The
Opal experiment uses a "4--th Dimension" data base on a MacIntosh for the
inventory of their electronic modules and find this very convenient with the drawback of
no easy access for the general members of the collaboration. The DD Online Computing group use Oracle to maintain the pool of online equipment
(several thousand items). This is now used by a wide range of people for arbitrary queries
and has been extended to include experiment hardware configurations of equipment not owned
by the group and also to include information on maintenance contracts. This runs on the
IBM VM system and makes use of the Oracle report generator. They did not consider using a
PC based system because of lack of connectivity with other central data bases. They feel
there is a gap between the very competent technical support offered by the DD data base
section and the end user community (indeed the User Support group feels they would need
dedicated Oracle experts to do other than refer users to this section). An interesting data base application is the one set up by the DD Fastbus System
Management team which allows experiments to generate code which is used to initialize a
Fastbus configuration (this contains full configuration and routing tables, arbitration
levels and so on). Fastbus is a multi--level triggering system where the hardware is fully
addressable and functionally defined. By using an Entity--Relationship model they showed
the data describing a Fastbus configuration to be relational in nature. They thus use an
Oracle data base on the central VAX cluster to store the basic descriptions of the
different Fastbus modules and the relations between them. A given experiment uses a set of
Oracle SQL forms (some tens of thousands of lines of SQL*FORMS containing many
inter--table checks) to build the description of their particular configuration. A Pascal
program then reads the resulting data base and generates the required configuration
initialization code in an ordinary VAX file which can be down--loaded to the experiment
concerned. Use of Oracle gave them the benefits of a backed--up data base with a
high--level of integrity, possibilities of roll--back and multiple concurrent access. The most striking fact appearing in the answers from the experiments is the large
variety of solutions adopted. Several experiments expressed their dissatisfaction about
the chosen solution. Such a situation seems to have originated from the regrettable lack
of general purpose data base applications software adapted to their problems. The design
and development of such general purpose applications needs expertise with a given DBMS and
manpower. A simple ad hoc solution is often "cheaper" for the individual
experiments, but, summed over the experiments, it is certainly more manpower consuming
than to provide once a general package. Some exceptions should be noted, for instance the
NADIR system developed in the Aleph collaboration. It is also striking that this package
was adopted by other collaborations, like Delphi, although it needed some
"customizing" before it could be used. NADIR is a good example of an application
very close to being a general solution for mailing lists, which is now being worked on by
the MIS unit. This example shows that when a sufficiently general application is provided,
the experiments will tend to use it, independent of the complication of the underlying
DBMS. A general observation is that experiments are using both commercial relational data
bases and the traditional HEP data management packages, depending on the nature of the
data concerned. Some translate the relational data base into a HEP package for reasons of
portability. Several areas can be identified where the use of a DBMS is clearly justified. and is
quite similar for the different experiments. Each of these areas has specific requirements
and needs different application software for access and update by the users. Some
applications already exist, most of them need, however, further development. Even where
this may require a sizeable effort, overall harmonization seems to be the only way to
avoid the present wasteful approach of multiple ad--hoc solutions, each of which is
manpower--intensive. We will first discuss the fields where a HEP--wide solution can easily be envisaged.
For these applications a complete tool package should be written by some DB
"professionals" in consultation with an advisory team" from different
experiments. This is then followed by a short discussion of data base applications where a general
purpose package seems to be - at least at the moment - not in sight. We believe, however,
that at least for the modelling phase and for the transportation of data bases general
tools can and should be provided. As mentioned above this type of data base should contain all data necessary for the
information exchange (mail AND electronics mail) between all members of a collaboration
such as address, phone number, office, function, responsibility, affiliation, network
addresses, etc. It will contain only few, well defined, tables. To avoid unnecessary
duplications and work this data base should be coupled to already existing ones like: user
registration, telephone directory and personal data. All members of a collaboration need read access to (part of) the data base via panels
and/or via coupling to high level language programs. However, since probably only reduced
functionality is needed at some places the access can be realized on an extracted d/a file
or even a sequential file. All other access might be (should be?) limited to the
secretaries of each collaboration. At least part of the data base should be accessible HEP
wide. It would be desirable if each member of a collaboration could initiate the changes to
the DB as far as his own data are concerned. In practice this would essentially mean that
a panel system has to be provided which creates the necessary update corrections sent then
to the Data Base Administrators (e.g. the group secretary). Since probably only reduced
functionality is needed at some places the access can then be realized on an extracted d/a
file or even a sequential file. Since the requirements are almost the same for all experiments, a general data base
package including all necessary tools for the communication with the data base and for the
transport to/from external institutes should be provided by CERN. As a starting point the
NADIR system could be considered. This application can best be realized on top of a relational data base. If done so it
will be quite easy to increase the functionality with time, whenever needed, without
having to redesign the already existing parts of the data base. Tools are needed for: With this database we want to keep track of all electronics modules and materials used
inside a collaboration. Therefore data such as module description, address of
manufacturer, purchase information, inventory and location are entered. The requirements are quite similar to the ones mentioned for the experimental
administrative database. In addition coupling to a general data base describing the
internal functionality of modules should be envisaged. This data base will be accessed only by few members of a collaboration. Therefore it
can be kept centrally at CERN with some access from outside institutes. The package itself
should ,however, made available to all collaborating institutes to have their local
bookkeeping system identical to the one used centrally. As for the experiment directory this is an important area where a general product
should be written and maintained. In a large experiment it is difficult to follow the history of a single event
throughout the different phases in the off--line analysis. Therefore a database is needed
which keeps track of files and program versions used for real and MC--events. In addition
a database which contains information on storage resources (tapes, optical disks etc.) is
needed. This data base has a simple and well defined structure and could be realized on almost
all DBMSs. What is important here is that read and write access is needed from terminals
and from higher level programs (e.g. reconstruction programs) in real time. It has to be
accessible - may be not with the same functionality - from all institutions of a
collaboration. As for the previous application areas the structure is well defined and a general
package can easily be designed. If realized on top of a RDMS, experiment dependent
additional information can easily be integrated. In a complete detector description database the geometrical description of the detector
components, the description of the electronics path to each sensing device, the
deficiencies, the alignment corrections and their variation with time have to be entered. In general, this leads to a complicated structure with many different tables
(100--1000). To cope with this complexity of the data structure and to allow for easy
navigation through it a RDBMS is needed. In particular, there are basic records and,
separately, update records which are merged with the basic records only at the time of
access. Access to the data base is needed by many different applications and packages in read
and write mode. Access time is crucial, hence the data base is probably read in once and
all access is made to a structure in memory as long as there are no changes in the time
dependent constants. In this complex system updates to the data base have to be done under the control of
the Data Base Administrator of the collaboration. The data base has to be available (with reduced functionality) on all computer systems
at CERN and in the outside institutions. Unless one introduces strong restrictions a general package cannot easily be provided.
However, it would be desirable to model an example data base which could be used as a
starting point for experiments which have a small number of (simple) detector components. In addition to the tools already mentioned for the experiment directory, more and
probably application dependent tools are needed, especially if the commercial DBMS cannot
be accessed directly at all places. Calibration constants used for (online) corrections of experimental data present
generally a huge amount of unstructured data. Therefore, for most experiments there is no
strong need to store them directly into a data base. However, the data have to be
accessible via the database. It would be sufficient to enter the access information into
the data base and keep the data on separate files outside (but nevertheless as part of)
the data base especially since some coupling to the detector description data base is
needed. The access to the data base is almost exclusively done from higher level language
programs. General recommendations are presented first. They are followed by further
recommendations specific to certain areas. As mentioned at several occasions in this report, a large duplication of effort has
gone in developing real or pseudo "data base applications" inside the
experiments. It frequently happened that a simple ad--hoc approach was chosen, which in
the long term proved to be too restrictive, once the volume of data increased or the data
base was needed for applications it was not originally designed for. It also happened that
commercial systems were tried and found too slow to be acceptable. The most frequent
arguments used by opponents of data bases remain: Specific applications should be written as part of the general support software. This
should be the result of a concerted effort between the DD division and the users.
Suggestions are given in the specific recommendations below. Although DBMS have found a wide acceptance in the engineering world (e.g. LEP), there
is still a reluctance for their use by physicists, who often prefer their
"homemade" solution. It should however be realized that, in the future,
experimental data are more likely to be managed by a DBMS than by our existing packages,
which remain tightly linked to our prejudices on Fortran as the main language used for our
programs. We recommend that: " An effort should be made to make physicists in
experiments more aware of the potentialities of commercial DBMS for their applications.
This could be achieved by intensifying training in the areas of data models (software
engineering) and DBMS. " It should be avoided that the effort to write General Purpose or applications code be
duplicated by the different collaborations, as is often the case today. To help overcoming
the initial difficulties and the long term maintenance, we recommend that: "
Manpower should be made available to support centrally the experiments, starting with the
design of the data base and continuing during the whole life cycle, including the
implementation of the application dependent code. This support team should also ensure the
long term maintenance of the General Purpose applications described below. " As the design of a data base application involves software methodologies and as
software methodologies make implicit use of data bases, the same support team could cover
both areas. This will help to speed up the implementation and to ensure that the resulting
application can be used by several experiments. It should be realized that the usage of a DBMS for a specific application requires one
to several man--years of work to design the structure of the data and to write the
necessary SQL and/or SQL*FORMS code. Several areas described below will need such
application code to be written. It has been stressed at several occasions that this task
is greatly helped by the use of a formal Data Model. This immediately provides a link with
the "Software engineering group", but we independently recommend: " A
package should be provided to design interactively a Data Model and to store this
definition in the form of a dictionary in ZEBRA files. The Entity--Relationship Model and
related software from ADAMO should be considered as a first step in this direction. This
would allow to profit from the experience gained and possibly from existing tools,
including commercial ones. " The Oracle Relational Data Base Management System has found a fairly general acceptance
in the user community as the standard data base system. It is already being used and will
be used even more in many different applications. Some features which were stressed by
users at several occasions are its handling of concurrent write access, its user interface
via SQL*FORMS and the possibility to implement tools on SQL to guarantee the data
integrity. A serious problem with Oracle is its price, which prevents it from being used
throughout the HEP community. As an indicative value, it costs approximately 25 kSF for a
VAX780 and 10% per year for maintenance. Other tools have to be paid for in addition. A
solution must be found which allows also the smaller institutes to use the data. "
A package should be provided to map data from Oracle to a ZEBRA (RZ) data structure. The
reverse could also be implemented, provided a data model describes the structure of the
data in the DB. A decent user interface should be written on top of these files to allow
the users to inquire about the information contained in this structure and to update it.
Tools provided with the ADAMO package could be used to learn from the existing experience
and could possibly be used directly as part of the proposed package. " This approach guarantees that the information can be used and updated in the
collaborating institutes free of charge. Several data base applications, all using Oracle, have been developed, like the User
Registration DB, the EMDIR system, several mailing lists from experiments using NADIR and
the AMSEND electronic mail system. Data are often duplicated and contradictory, because
some are out of date. It is recommended that: " A data base should be set up
covering all CERN (or HEP?) users and other people related to experiments. It should link
with information from existing data bases. It should include the functionality required
for experiment mailing lists and experiment specific data. Control of the data, i.e.
entering and updating the information, should stay within the experiment concerned. We
further recommend that a study be made on existing tools and their performance, in order
to coordinate any future efforts, such as those being made around NADIR and EMDIR. The
functionality should cover at least the one of the NADIR and AMSEND systems. " LEP experiments seem not to have invested much manpower in this area yet and a common
solution still seems possible. Therefore, we recommend: " A solution should be
researched and developed urgently, in common between the LEP experiments, in the area of
tape bookkeeping, to avoid duplication of effort. " The part of the software description related to program code and data description may
come automatically with the Data Model tools and is considered as part of recommendation
3. In addition, "paper documentation" will still be required and makes this an
important area to be looked into. CERNDOC seems to have basically the right functionality,
but it should be coupled to other data bases, like the CERN Library data base and
experiments administrative data bases. It is also essential that the data can be
transported and printed throughout the HEP community. We recommend that: " The
redesign of existing documentation data bases (CERNDOC, HEPPI, ISIS etc.) into a common
data base system (e.g. Oracle) should be envisaged. " This is probably the area with the largest investment of manpower and the largest
savings if a common solution could be found. What makes it particularly difficult is the
large data volumes encountered and the high speed needed for accessing the data. It seems
unlikely that the existing choices for the detector description and calibration constants
data bases could be modified, given the imminence of the data taking for the experiments
involved. It also seems difficult to provide a completely general solution, given the
large differences between the detectors. It is at least expected that the experiments will store the information in a DBMS and
in ZEBRA files. The availability of software to handle a data model and to convert between
a DBMS and ZEBRA could give some help. Some guidelines and assistance for structuring the
data would also be helpful. The experience gained with the usage of the large data bases
set up by the LEP (and other) experiments should be exploited to try to define further
support tools and minimize duplication for post--LEP experiments. A study group should be
set up around the end of 1990, after a sufficient experience is accumulated. PAW datasets are expected to play this role. PAW is already widely accepted by the user
community, and no new software needs to be developed. The design of PAW explicitely does
foresee do use PAW datasets as data base; the present PAW implementation, however, seems
to be lacking the built--in possibilities for identifying the contents of the various
directories. In--line identification seems necessary to make PAW data exchange inside and
between experiments largely self--documenting. Databases and bookkeeping for HEP experiments, ECFA Working Group 11, ECFA/83/78 (Sept.
83). This subject was considered by a separate working group. To define what is meant by
software engineering: 'the application of engineering--like methods for large scale
software production'. In this report we consider the methods and tools which are likely to
become a necessity for such complex projects as will be experiments (and accelerator
control) at a future hadron collider. The major recent trend in HEP experiments and one that promises to continue, is that
experiments have become very large, in terms of equipment, numbers of physicists and data
volumes. Experiments, in addition, have lifetimes measured more in decades than years.
Necessarily, the software, both on--line and off--line, required to analyse and control
these projects has also grown in complexity and volume. The issues of how to build, manage
and maintain large software projects are addressed by modern software engineering (SE)
methods and techniques. Furthermore, SE methods will undoubtedly be essential to enable
the integration of both hardware and software in future real--time systems. The question of which programming language(s) should be used is not discussed by this
working group. It should be stressed however, that interesting developments in such areas
as automatic code generation and the like are not likely to be commercially provided for a
Fortran environment with typical HEP memory managers and associated data structures. We have no intention of recommending the precise tools and methods that should be used.
This is a field which is rapidly evolving and it would be presumptuous to presuppose what
might exist in 5 years time. We do, however, base many of our conclusions on the
Structured Analysis/ Structured Design (SASD) methodology that, thanks to pioneering work
by the Aleph collaboration, has gained a strong foothold in the HEP world, both at CERN
and at other HEP laboratories. In the following, we consider in some detail analysis and design methods and related
tools which constitute the relatively recent major innovation in software engineering. We
also discuss briefly the tools required at the coding and implementation level. This
includes code management which, traditionally, has been the most widely used
implementation tool. Modern analysis and design methodologies provide systematic techniques for the analysis
of a problem or project, together with design strategies for translation into software.
The basic components of a typical method are:
In addition to the above, 'reverse' engineering tools are used to analyse code and
produce diagrammatical descriptions in order to verify what the code really does. Given
the addiction of HEP to Fortran and other concepts such as memory managers, these tools
have, in general, to be home grown. The term 'software crisis' was coined in the 1960s as recognition that large software
and system development was, in many instances, out of control. The various studies made of
large projects indicated that a major contributor to this problem was the lack of a
systematic approach to software design and development. The concept of 'software
engineering' emerged with the idea of applying engineering--like techniques to the various
stages of the software life cycle. With the advent of significant new ideas on Structured
Design and Structured Analysis [Footnote: Some literature: W.P. Stevens, G.J.Meyers, L.L.
Constantine, Structured Design , IBM Systems Journal 1974; T. DeMarco, Structured Analysis
and System Specification, Yourdon 1978; C. Gane, T. Sarson,Structured Systems Analysis,
Prentice--Hall, 1979; G.J. Myers, Composite/Structured Design, Van Nostrand Reinhold,
1978; E. Yourdon, L.L. Constantine, Structured Design, Prentice--Hall, 1979. ] in the
early 1970s together with the concept of the software life cycle, it became possible to
formulate frameworks that incorporate the known techniques for software production and the
appropriate management practices. Whilst the concepts of software engineering have been embraced in particular by the
aerospace and oil industries over that past decade, the start of any significant
involvement in the HEP world can be traced to 1984. The workshop on "Software in High
Energy Physics" in 1982 provided the HEP community with a chance to meet and discuss
with people who not only advocated but had actually used these techniques. Several groups,
notably the Aleph software group, expressed further interest and, with the aid of the
results of several comparative studies Aleph selected SASD as a suitable methodology. This
selection was based on the criteria that SASD supported all phases of the life--cycle, was
widely used in industry, was relatively easy to learn and use and was itself supported by
(at that time rather limited) automated computer tools. This view was confirmed by a
tutorial on software design techniques in 1983. Subsequently courses in SASD were given at CERN and towards the end of 1984 SASD was
formally adopted by Aleph for both the on--line and off--line software. In the following summary of the existing situation, SA refers to the analysis part of
SASD and SD to the design part.
There is currently no official CERN support for software engineering analysis and
design methods and tools. Partly as a result of the Aleph experiences, the FNAL experiment D0 adopted the same
methodology for its software development. Subsequently FNAL has negotiated a license for
use of a modern commercial system based on SASD. The first users of this tool are D0 and
the FNAL data acquisition group. Consultancy support is eventually foreseen. Towards the end of 1986 and following into 1987 courses on SASD were given to
experimenters from the two HERA experiments, the HERA machine group and the DESY computer
centre. It was decided to adopt SASD for both on--line (ZEUS) and off--line (ZEUS and H1).
In November 1987, after evaluation and comparison of existing products, a common analysis
and design tool [Footnote: CADRE/Teamwork, by Cadre Technologies Inc, Providence, R.I.
This tool has been selected by (amongst others) : FNAL, DESY for HERA, Cornell for CLEO
and CERN for SPS controls. ] was selected, with licences negotiated by DESY and provided
for the experimental groups. The ADAMO toolset is used. In this chapter we attempt to summarize the experiences of large scale software
production both with and without the use of software engineering methods. For the former
we have looked at major experiments including Delphi, EMC, L3, Opal, UA1 and UA2. The
software for the older experiments (EMC, UA1 and UA2) was written well before SE
techniques became fashionable. Delphi and Opal use SE techniques to some degree, but the
methods have not been officially adopted by the collaborations. In addition we reflect on
the SPS controls software prior to 1985. The conclusions on the use of SE techniques are based on the experiences of Aleph and,
to a lesser extent those of ZEUS and H1 at HERA and D0 at FNAL, together with those of the
SPS and LEP control systems. Experiences at the European Space Agency have provided
specific insights into project management possibilities. The conclusions and experiences below are an average and not necessarily applicable to
all projects. Reliance is made on dynamic memory and data managers to organize data structures. The
latter are in general well documented, both with text and (often) diagrams, and contain
some elements of data dictionaries. Access to data banks is sometimes provided via a
higher level package, which thus acts as a sort of data model. There are clear analogies
with the SE principles outlined above. There exist general problems that do not specifically result from the lack of adoption
of SE methods ; rather they reflect the attitude of the management of the project that
software building is a second rate activity compared to, say, building detectors. This
often results in no formal managerial structure for the software team which itself is
often largely composed of young physicists who have little or no experience working on
large software projects. Furthermore there is usually no means of obtaining suitable
training. Reliance is made on consultancy provided by in--house experts who often
themselves have received no formal training. One of the main reasons why SE has not had a bigger impact on the more recent
experiments is the lack of money for automated tools and, until recently, lack of good
tools themselves. The following is a summary of other problems which are addressed by SE methods and
techniques.
It should be emphasized from the outset that SE techniques are not a panacea. Rather,
the use of these methods brings with it an awareness of the necessity for proper
managerial structure and proper training. The adoption of a particular computer aided
tool, however versatile and powerful, will do little to solve the problems of large scale
software production without the necessary commitment to apply the systematic
engineering--like concepts inherent in SE methodologies. For SE to be effective, the project management must understand the principles, support
the effort and be prepared to make serious commitments in money, for the necessary
hardware and software tools, in time, for training and learning, and in manpower at the
start of the project. The methodologies comprise a wide selection of techniques and practices, and it is up
to each project team to select the sub--set they feel comfortable with. Though the HEP
environment differs from that in industry in terms of management structures, it is clear
that it should be as unthinkable to work on a complex software system without following
guidelines and having some level of expertise as it would for the non--expert to take a
soldering iron to the inside of a Fastbus module in the hope of improving it. To cope with the problem of high learning restrictions before physicists can start to
work on large software projects, it should be recognized that not every one needs to be
equally skilled. Alternatively, in this domain of specialized skills, a certain division
of labour is required. In the SPS controls software system upgrade the results have been quite positive with
relatively straight forward design and coding following a proper analysis phase. The
system performs well with very few changes needed in the requirements specification. Use of SE methods works both in a traditional HEP environment, where there is a broad
overlap between analysts/designers and those writing the code, as well as a more
structured situation, such as LEP controls, where accelerator physicists analyse the
requirements and professional programmers perform program design and implementation. The following is a summary of more specific conclusions:
To summarize, software engineering techniques have been successfully used in Aleph,
accelerator controls and other CERN projects. The methods have been very helpful in
providing a clear understanding of the problem at hand and as a means of communication of
ideas within the groups both between experts and non--experts. Similar conclusions have
been drawn by groups from FNAL, DESY and other HEP laboratories. In the past, the lack of
suitable automated tools has been a restriction. This situation has now changed. Good
tools are available and further improvements can be expected. The majority, if not all, of
the groups mentioned in this report have selected the same SASD based tool (Teamwork)
which is therefore becoming the de facto HEP standard. Software development does not require full--time presence at CERN and is therefore an
area in which physicists in the outside institutes contribute to experiments. This is
especially true of small groups in universities which lack the technical resources to make
contributions to detector hardware. Some of these groups have local computing resources,
such as medium sized mainframe computers or workstations, others rely on network access to
remote computer centres at CERN or elsewhere. Some institutes have quite limited resources
for buying new computer equipment and software. Within the above constraints, the users in institutes outside CERN have much to gain
from SE techniques. The tools for breaking large problems down into well defined parts
which can be tackled independently are of particular benefit to the community outside
CERN. Such subproblems can be taken on by individuals or small groups of users in the
outside institutes. While the work is done independently, the common use of SE tools
ensures consistency, so that the code which is produced should fit together correctly. At
the level of a common data dictionary the different groups may maintain close
communication. The diagrammatic tools of SE, such as data flow diagrams may also improve
communication between groups at remote sites. However, as already noted above, SE tools
will not remove the need for discussion, and regular meetings between contributors to a
software project will continue to be important. Training in SE methods and tools must be provided for users from outside CERN. Where
possible this is probably best done nationally to save on substantial travel costs; this
is easiest in countries with a large national laboratory, e.g. Rutherford Appleton
Laboratory in the U.K.. In any case, courses generally only cater for a limited number or
people (15 or so) for maximum benefit. Good documentation is clearly a necessity - both
pedagogical and reference. The special problems of training students need to be
considered. In general it would be sufficient that limited training be given so that they
could use the results (e.g. of analysis and design) of SE techniques. Standardization between the different HEP laboratories would be highly desirable since
many institutes are involved in collaborations with more than one laboratory.
Standardization with other users of national laboratories would also be useful, although
probably harder to arrange. The benefit of this to the outside institutes is that costs of
hardware, software licenses and expert support may be shared. A more general benefit is
that physicists will not have to learn new tools when they change experiments. At present, some of the best SE tools are available only on workstations which are
still relatively expensive items; the software licenses can also be very expensive. Before
committing the CERN community to a particular computer aided software engineering (CASE)
package, the financial implications for the outside institutes must be fully considered.
It would not be acceptable for HEP groups outside CERN to be excluded from contributing to
the software effort due to the choice of an unnecessarily expensive set of tools. In view
of this, it may be worth considering systems which are available at relatively low cost
and which run on cheap or existing hardware. The problems of maintaining a central database for SE tools will have to be faced. One
of the key ideas in (some) SE methodologies is the data dictionary. This at least must be
maintained centrally, although copies may be made for reference. The problems of
maintaining such a database, with updates coming over wide area networks, must be
addressed. The software engineering environment is completed by implementation and coding tools.
In this area tools have already been widely used in the past, and experience exists. We
present here a brief review of the more important tools and their functionality. No
mention is made of standard compilers, and we also exclude Documentation and Data Base
system tools, discussed in other parts of this report. Source code management means the use of a tool which helps to maintain different
versions of a program (or other text file). Various tools have been in use by most
experiments over many years. Several exist today and have shaped the habits of their
respective user communities. No tool exists which clearly responds to the modern
interactive workstation--oriented development environment. An overview of the situation in the HEP community is the following:
There is no consensus concerning source code management requirements in the community.
In view of the rapidly changing environment, new projects are not recommended in the area
of code management tools. In the long term, it is possible that Unix will offer a
sufficiently portable and friendly environment which may obviate the need for portable
tools. Opinions differ considerably as to when and if this will happen. In addition to the functionality traditionally offered by source code managers, other
important tools exist that help in the development and maintenance of software projects.
The following is a partial list of some of the more useful categories.
An important aspect in the choice of support tools is the selection of a coherent set
within a common framework. There exist several ambitious projects [Footnote: The Toolpack
Project from Numerical Algorithms group and PCTE : Portable Common Tool Environment
(ESPRIT). ] that attempt to create a global software environment consisting of an
integrated suite of software utilities. In order to improve efficiency of software production, programmers must change their
habits to make increased use of such tools. This will require both learning time and
financial investment. Effort is required to keep abreast of new developments in this
rapidly changing field and also to provide local support for selected products. Large software production is a serious activity and deserves, from the start of the
project, as much attention as detector building. Adequate software tools are required as
well as the necessary computer hardware. " Software engineering tools and
methodologies have been shown to be effective provided sufficient support is given by the
project management. Use of such tools alone will solve little, unless the required
managerial structures and practices are adopted such that very large collaborative efforts
may be partitioned into smaller teams working in parallel. " Our choice is "pay now or pay lots more later". [Footnote: From: J. Manzo, On
Managing Large Scale Projects : Some Simple Principles for Developing Complex Systems
,Comp. Phys. Comm. 45 (1987), 215-228 ] In this context later could mean never or at best
only partially fulfilled physics goals from experiments employing vast manpower and
financial resources. " Training in the concepts and ideas of software engineering is essential and
should be provided by (repeated) courses given by experts from HEP. "
Realistically it should be recognized that, given the range of skills needed to design and
construct large scale software, there has to be some division of labour within the project
team. Different skills are required by the project managers, say, than by those working
more at the implementation end. A central compilation of the documented experiences from past and existing projects
would provide the necessary complement and back--up to formal training. Support and consultancy is required on several, not necessarily independent levels.
" A central support group should be set up to evaluate European HEP requirements
and negotiate licences for the necessary tools. This same group would also act to provide
feedback to and from the vendor for bugs, problems and improvement, and help with product
installation. " This group should coordinate the use of SE tools across different
projects and different HEP laboratories, and build and maintain extensions required by the
HEP community. " A special interest group should be formed consisting of users of
SE tools both at CERN and the outside institutes. " The functions of this body
would be to provide communications between the groups involved, to decide on the common
tools required and to make sure that where possible, progress may be made on common
problems. " Day to day consulting on conceptual problems is best dealt with by
expertise from the project itself. This requires an establishment of a base of expertise
in each project. " Automated computer tools are essential. " A powerful set of tools capable of
supporting development of a variety of projects (batch oriented systems, real--time
applications and data base modelling) should be selected with a view to standardization
across the CERN and HEP community. The relevant hardware environment is the workstation.
" Adequate, though not perfect, tools exist today. It should be remembered that like any
other tool, they will evolve, and new ones will have to be obtained as requirements
change. " CERN should negotiate site--wide or HEP--wide licences for groups working on
experiments at CERN, and for other required applications. " Software development will remain a strong, major activity of the outside institutions.
When negotiating licences it will be important to keep in mind the limitations of hardware
and financial resources of these institutes. Standardization of SE tools, both CERN--wide
and HEP--wide is very relevant in this context. Training for outside users should be
provided at CERN although not exclusively. The subject of graphics support was considered by a separate working group. We
gratefully acknowledge the kind collaboration of P.Schilling and J.Olsson from DESY. Within the last years the range of graphics applications within HEP has become
extremely large, from the simple display of histogrammed data to highly interactive
dynamic event viewing. This growth in the kinds of application programs has been
paralleled by the very fast development of graphics devices and workstations. The types of
applications considered in this document are as follows:
With respect to the above list of applications we will concentrate essentially on
software development and experimental analysis which will be, together with the
publication and human interface area, the main use of graphics during the coming years of
the LEP era. The large range of devices available from numerous manufacturers (all with different
native graphics interfaces), the large number of graphics applications mentioned above,
the need to run programs on several types of computers (workstations and mainframes)
inter--connected through networks, and the large number of physicists participating in the
current and future experiments, are all factors which lead to the following requirements:
It is essential that a common approach to graphics and user interfaces is taken by
programmers in different areas so that modules of code will work together in a compatible
fashion. The different layers of HEP graphics applications can be classified in three levels:
The first 2 of these levels will be discussed at some length as the fundamental
building blocks of all graphics applications systems. " Not only must applications be able to work at any one time on a range of graphics
devices, but over the time--scale of a LEP experiment it will also be necessary to adapt
to several generations of equipment. Whatever choices are made for hardware and software standards for the first generation
of LEP experiments, they must take into account the need for a smooth evolution of the
applications programs. " The large number of graphics packages (Application Program Interfaces, or APIs) like
WAND, GPR, GSR, Template, DI3000 etc. cover the full range of possible hardware in a
non--transportable way. Thus, since it is not possible to standardize application programs
to only one type of graphics machine, the 'natural' solution has to be provided by
graphics standards. (Ref: Report on the Status of Graphics Standards, D.R. Myers,
DD 88--14.) Of course, if use is made of features in a standard, which are not
supported in hardware and hence have to be emulated, then the performance will suffer.
Nevertheless, as an example, the software emulation of 3D displays, although slow on
low--end graphics machines, does allow the development of 3D code on cheap hardware, and
may also be sufficient for applications which require 3D but do not need real--time
transformations. As a minimum, the features required of a standard include the use of hardware
primitives, characters, segments and real time transformations (when available). However,
at the top end of the scale, a particular device may contain features not supported by the
standards. It should be noted that the current standards, such as GKS and PHIGS, provide
inquiry functions to allow one to find out at run time which features are available, and
thus to decide which algorithm is best suited to the occasion. Whilst using a graphics standard, such as GKS, for all new graphics devices can result
in a significant loss of performance compared with what could be attainable utilizing the
hardware features directly from the application program, to a large extent performance
depends on how much effort is available when writing the device drivers. Performance is
certainly the reason why some people using workstations prefer the native graphics package
to any standard. However, there have been several notable cases where poor performance was
found to be due to the use of a standard in a completely inappropriate way. " Training and experience in the use of GKS (or any other standard) is essential to
produce well written and efficient graphics applications, and CERN support in this area is
necessary. " In the more complicated examples of HEP application programs, for example event viewing
and analysis programs, the complex data structures within the application program are
represented in the structures within the application code, for example in Zebra or Bos
banks. Since a dynamic hierachy of coordinate systems is not required it is sufficient to
map the physics data structure to a flat graphics structure, such as a set of GKS
segments. This also avoids the complexity of programming graphics data structure editing
functions. GKS does not provide explicit support for workstation 'window' systems. However,
assuming that the workstation operating system provides a window manager to manipulate the
windows, GKS does allow multiple windows to be used, with some limitations, and this
feature has been implemented at the driver level for several existing workstations. "
From the experience gained so far, GKS provides a very good standard low--level
graphics application interface, giving a reasonable compromise between portability and
performance. " However, as GKS is by definition a 'kernel' system, it is recommended that a graphics
library of higher--level functions should be developed, based on what has been done in
various experiment over the previous two years. The working party believes that the
supported basic library should be layered on top of GKS, and this can be used as a
starting point for those who wish to adapt the library to non--standard picture generation
software package. This library will include dedicated HEP primitives, some of which are
described as graphic macro--primitives in the Higz manual, as well as features such as
histogram production, helix drawing, jet drawing, etc. This would help to prepare a smooth
transition of the existing code toward new hardware and graphics standards which might be
used in future experiments. " We recommend that a set of high--level graphics tools and functions based on GKS of
particular use in HEP should be provided as part of the CERN Program Library. They may be
written by groups or individuals throughout HEP, but the work should be coordinated at
CERN, which should also provide long--term support. " Interactive graphics will run increasingly in a distributed environment, hence it is
essential that we gain practical experience in X--Windows implementations, and thus be
able to give sensible advice to those in the community wishing to acquire software
products to support distributed graphics. As X--Windows is a very low--level protocol, it
will also be necessary to evaluate the various tools available layered on top (DECWindows,
Open Dialog, etc.) and to choose one, or at most two of these as the CERN supported HEP
interface to window systems. Moreover, X--Windows is not a graphics package, so an
important area to explore will be the co--existence of X--Windows with layered graphics
packages (perhaps using the X--Windows extension called PEX). Also we have to solve the problem of communication from local interactive graphics
programs with remote databases containing thousands of histograms, DSTs and raw data, as
well as to define a framework open to the evolution of the hardware to new systems (Unix)
and to new languages (C, PROLOG...). " We believe that networked graphics, where all or part of a graphics application runs in
a host remote from the graphics workstation, will be of increasing importance during the
period addressed by this report. We recommend that CERN effort should be put into gaining
experience of windowing implementations, with a view to selecting one or two as supported
interfaces for the HEP community. " With the introduction of the CRAY running UNICOS, Unix has made a definite appearance
in HEP, and we expect to see expanding use of this operating system. The CERN libraries
are supported already on Unix, and one can forsee the use of other languages which
complement Fortran in the graphics area. The use of Unix would certainly ease the use and
support of a wider range of workstations, including those from SUN, Silicon Graphics,
Ardent, etc. Our goal is to provide more choice via the use of open systems. Another area which is still developing is the integration of graphics with documents.
The signs are that CGM will emerging as the accepted picture storage medium for such
applications. However as many hard--copy devices manufacturers are providing built--in
Postscript interfaces, we should not ignore this de--facto standard. In addition, the
'appendix E 2D metafile format' of GKS is now an acceptable standard, but the lack of
useful 3D Metafile standard must be corrected. It is acknowledged that many of the major institutes in the large collaborations plan
to use at least one 3D workstation and several 2D workstations in addition to normal
terminal utilization for software development and experimental analysis. The range of possible graphics hardware is very wide, and is evolving at a rate much
faster than the progress in the preparation and analysis of experiments and the lifetime
of software packages. Applications exist needing high--resolution, 2D and 3D, colours or
grey levels (sometimes with a palette of 256 colours for smooth shading), and
high--resolution hard copy facilities. For the coming LEP experiments wire--frame
representations seem to be adequate, but we believe that for future experiments (e.g. at
LHC) we may well be interested in solid modeling techniques. Nevertheless, there are many
less--sophisticated applications, and so hardware with a range of performance levels is
required. In most cases we need full interaction with the screen, both to change
dynamically the aspect of the graphical representation, and to select physics quantities
by 'picking' their graphical representations. At both ends of the performance range there is a need to use simple graphics terminals
on mainframes for program development, as well as 3D graphics machines with hardware
processors to do real time transformations. Although the trend is clearly to replace
terminals by workstations, due to the life--time of the range of machines currently
existing or being bought, terminals will still be in use during the period covered by this
report. We have also found that workstations currently available with a typical
performance of one MIPS are not sufficiently powerful, either for compilation and linking
of large programs, or for good graphics performance. Fortunately, the trend continues to
be towards faster machines, and affordable devices are appearing with 4 to 10 MIPS. This
is already sufficient for reasonable 3D performance for many applications which do not
need smooth transformations in real--time. At the high--end, CPUs are approaching the speed necessary to perform dynamic rotations
of 3D wire frame objects without special transformation units. This has interesting
implications for the utility of display lists, as their usefulness is less obvious for
machines at this level. Although ultra high--level graphics performance is becoming
available in the some workstations, we are more interested in enhanced CPU power, since
our work concerns CPU intensive interactive computing using complex data structures rather
than computer design applications. Already in the 'MUSCLE' report, as well as in another chapter of the present work,
networking between mainframes and workstations or clusters of workstations has been
identified as being an important issue. We certainly endorse that conclusion. Tools which
optimize file transfer and the transmission of graphics commands over heterogeneous
networks will be of the utmost importance for graphics applications. Whilst it might be
necessary to group workstations (perhaps of a similar type) in clusters with a database
server, the ideal is clearly to be able to work in a heterogeneous environment with the
minimum number of constraints. " We recommend that CERN coordinate the definition of application--oriented benchmarks
for 2D and 3D devices, these being based on GKS 2D and 3D code for the basic picture
generation. In addition CERN should, in cooperation with the institutes, perform benchmark
evaluations. " We endorse the choice of GKS as the graphics standard for HEP for the coming LEP
experiments. GKS will exist and be developed for more than 5 years, and new hardware
capable of running PHIGS can certainly support drivers for GKS--3D, although these might
need to be written or commissioned. The use of GKS--3D has the important advantage that 2D
modules of code may be used within 3D applications without modification, and that
application programmers do not need to learn a second interface standard. However, during
the period under consideration it will be necessary to monitor developments in the field
of graphics standards, especially as a new generation of standards is scheduled to appear
during this time. " CERN should provide support for the basic graphics packages and be responsible to
ensure that they run on a range of equipment, as well as for distribution and
documentation. CERN should also evaluate and test new terminals and drivers. It is
essential that sufficient manpower with the necessary technical knowledge be made
available to fulfil this task. " Application programs used in big collaborations are written in several places using a
range of hardware. Once developed, these programs are used by physicists who are not
software experts. The support to maintain the applications and test them in new
environments must be provided by named people within the experimental groups. If these
people can be located at CERN, so much the better. " In order to facilitate the communication necessary for decision making we recommend
that an HEP Graphics Steering Group be formed with representation from CERN,
collaborations, institutes and major computer centres associated with HEP graphics
processing. It may be sensible to look for institute representation from country working
groups where they exist, since such groups cut across boundaries between collaborations in
a positive manner. " This topic was considered by a separate subgroup, the work of which has proceeded
primarily via discussions with working experimental physicists and programmers. There have
been no questionnaires distributed to the HEP community and no response has been received
from the implicit invitation to contribute in the CERN computer newsletter. CERN provides major central computing facilities for experimental physics. There is, in
general, a small overlap between the activities of the professional staff running the
central computing facility so their training should come from commercial sources. Often
CERN staff are involved in the development of system software in collaboration with the
manufacturers and this contact takes on the role of training. Similarly, clerical and
technical staff are encouraged to attend courses organized by the Technical Training
Committee. Having ascertained that the Education Service has no surplus manpower,
terminals, workstations or conference rooms, we have made no in depth study of this area. We found that the majority of physicists rely heavily on the documentation and working
examples of the libraries and program products which they use. Many have acquired new
skills through reading books, guides and manuals. There was a wide range of responses to the idea of receiving formal training. Some
physicists were prepared to attend a course which suited their needs; most supported the
idea that training should be available but found it difficult to fit it into their own
work schedules and a few suggested that if they needed to attend a course to use a package
then this alone was proof that the package and its documentation were inadequate. Most physicists drew attention to their need for greater clarity, simplification and
general improvements of the user guides, reference manuals and other documentation of the
libraries and program packages they use. Two detailed points were that working examples
should work and there should be more simple examples in the documentation of the major
packages currently in use at CERN. The online FIND and HELP facilities were generally appreciated as was the assistance
which may be obtained via the User Consultancy Office (immediate help with simple problems
and requests for information, distribution of user manuals and access to reference
manuals, and the transfer of difficult problems to the best available expert). There was
however some concern, expressed particularly by physicists making occasional visits to
CERN, that even online documentation may be out of date and examples do not work because
they are not fully maintained in the evolving CERN computing environment. In general, physicists working outside CERN were content to receive printed copies of
documentation from their experiment and from major CERN products by air mail, and
corrections by telefax. A few expressed a preference for a service, to deliver documents
and updated pages over the electronic mail networks. This is difficult to achieve because
the printing and publishing facilities on the Central CERN computers are incompatible with
those in the HEP community as a whole. Everybody, from the worst hackers to the most talented and prolific producers of the
routines and packages used by experimental physicists found it difficult and time
consuming to produce and maintain documentation in English. The experts, with their
familiarity and detailed knowledge of their working environment, also find it difficult to
appreciate the needs of the end user. " To respond to the expressed needs of producers and consumers we recommend
that a qualified and experienced technical writer should be recruited. " There are a large number of creative people at CERN it is not too surprising that it
has often been a long and arduous process to reach agreement on a standard such as common
procedures for the production of technical notes, papers and other forms of documentation.
However, given the considerable dissatisfaction with the documentation within experiments
and the manuals for many CERN packages, we would expect that the authors and managers of
code should approach any new initiatives on the structuring of documentation with an open
mind and a spirit of flexible co-operation on their part. Training courses for physics-computing lasting two to five days are given approximately
once a month at CERN. The requests to attend these courses exceeded the number of places,
especially for those courses with a large component of practical work using terminals and
workstations. We found several cases where physicists had not reached the point of applying to attend
a course. For example, there were very important activities at CERN which took place
during the course and the course was given some months before the new skills would be put
into use and the knowledge would not be retained. The pressure to produce results, for the
design and installation of an experiment or the publication of results at conferences,
inhibits attempts to take the time to learn new techniques which might lead to more
efficient working practices. The demand to attend courses already exceeds the capacity and may be assumed to rise if
there were better access and a choice of dates and extra subjects were introduced. The studies of data-bases, software engineering and graphics in this report identify
several areas where (200) experts within experiments should receive training and, with the
introduction of the personal workstation, many of the (2000) physicists at CERN will also
wish to learn new techniques. " Therefore we recommend that training courses and tutorials should be given
repeatedly " For subjects given by visitors to CERN, or for physicists attending commercially run
courses suited to CERN's requirements then the number of people receiving training will,
most likely, be determined by financial constraints. For topics in which there is already expertise at CERN, or a CERN based package, the
people writing or maintaining the package often prepare and present the courses. This does
give them some useful feedback on the package itself. Once a course on a topic becomes stable and mature, or the level of demand requires the
course is given many times a year, then the author's time is not spent most effectively in
repeating the course. Resources should be provided to maintain an appropriate repetition rate. The courses may be given by trained demonstrators, the course may be recorded and video
training introduced, and eventually a large component of the training may be given through
computer based learning techniques. As courses are given more frequently they may also become more modular. The tutor does not have to cram in all topics of interest to people with differing
levels of ability into the only one week course which will be given on that topic for two
to three years. Pre-requisite skills may be defined (for example the ability to use a workstation, a
knowledge of the languages used in practical work and units of related courses) and units
suitable for the technical expert in a collaboration may be missed by the end user
physicist. HEP regional centres should consider providing host facilities for their user community
to attend live courses given on CERN based products by visiting CERN staff. There was a strong feeling within the working groups of this report that resources
should be allocated to educate the HEP community to exploit the advances of software
engineering and computing technology. It has been difficult to persuade busy, self--motivated physicists that the investment
of their time in learning new techniques may lead to improved productivity (or faster
physics). Indeed, there are physicists who have made almost daily use of the keyboards of
teletypes, punch card machines and computer terminals for over a decade have not found the
time to learn to type efficiently. We have seen the start of the widespread introduction of personal workstations into
offices at CERN, we expect 1000's by 1992. Thousands of physicists will use the CERN based
Physics Analysis Workstation (PAW) package for interactive data analysis. The estimated rate of the introduction of workstations leads us to believe that there
will be one or two new users of workstations per day at CERN. The new user, and visitors to CERN, should be able to sit with a relative expert for 2
to 3 hours and learn the basic use of the workstation, for example screen management, the
help system, the file system and editing techniques. Other material could be divided into
modules such as an introduction to the use of interactive debugers and the available
software engineering tools; electronic mail, text processing and graphical data
presentation; and remote login (terminal emulation), job submission and file transfer.
" We recommend that there should be an introductory session for new users of
workstations. " There may be a demand for several such sessions each week. This has proved to be a very
efficient way for a physicist getting started within the workstation clusters of large
experiments but it is a workload which should not be carried by other physicists. The Physics Analysis Workstation project (PAW) is an attempt to provide interactive
data analysis tools across a wide range of mainframe computers, terminals and
workstations. The project is strongly supported by CERN based experiments and the software
has recently been generally released for user feedback. In short, PAW may become the
interactive data analysis tool for thousands of physicist over the next few years. Physicists outside CERN given the manuals of the PAW suite of programs (and told by
their local system manager that PAW is now available) have reported that it is difficult
to get started. Although the product itself could be easier to use, there is clearly scope
for improvements to the manuals. We repeat that the introduction of more simple examples
is a frequent request and the manuals of the PAW suite would benefit from the attention of
a technical writer. Courses given at CERN on the use of PAW have been oversubscribed and well attended. The
excess requests from one course may completely fill the next course in three to six months
time. The demand for an introduction to PAW is expected to rise as the LEP physicists move
from installation to data analysis and as PAW becomes more mature. The policies and priorities for funding within many institutes make it difficult for
many physicists involved in software development and data analysis to come to CERN purely
to attend a PAW course. " PAW is an ideal candidate for the production of video training cassettes at
CERN and for the courses with personal contact to be given by "demonstrators"
rather than the authors of the component sub-systems. " The experts within experiments are aware of the pioneering work of Aleph in the use of
software engineering techniques. Many recognize the shortcomings of traditional methods
but now wish to concentrate on the physics analysis aspects of their experiment rather
than start to make the significant investment to adopt formal methods within a
collaboration after a few hundred thousand lines of code have been written. It should not be left to the "next" experiments before some training in the
use of software engineering techniques is introduced at CERN. Young physicists should have
the opportunity to be trained to use these techniques as they will be used in industry or
future experiments where they may eventually find employment. New projects could be
undertaking using the techniques and packages which are available today; for example the
data processing experts in the computer centre and the Lep experiments could co-operate in
the design and implementation of their tape handling utilities. We considered the possibility that the adoption of formal specification methods would
allow physicists to concentrate on the development of algorithms and analysis and use
trained programmers to implement the code. This proved to be very unpopular when suggested
to working physicists. Training in data modelling would also lead to an awareness of the potential use of data
bases. We have found that there is a significant acceptance of SQL as a language with
which to make database enquiries for book-keeping activities (during detector
construction). High level physics analysis languages are emerging as the LEP experiments draw near to
data taking. These languages reduce the work to be done by a physicist analysing data but
an expert may be needed when features are missing from the language. They give many
advantages when comparisons of analyses are made as many cuts and definitions are defined
within the package. At present there use seems likely to be private to one experiment and
its own training resources. In the past such packages (like macro SUMX) did not gain a
widespread user acceptance. It is considered to be a strength of PAW that user actions can
be taken via interpreted Fortran files (via COMIS). We are entering an era where modern software working practices may be introduced into
HEP work. The techniques which physicists and programmers should learn are discussed
earlier in this report. Some subjects may be taught to a large audience in the CERN
lecture theatres using traditional classroom techniques. However the bit mapped personal
workstation provides the best user interface to the software tools which will become part
of everyday working life. A large proportion of training activities should take place
where the physicists may learn through relevant exercises on workstations. " We recommend that the provision and operation of workstations and terminals
for hands on training and demonstrations should be a central responsibility. " A 'divisional workstation training office' in the computer centre has recently been
established. This has five Apollo workstations, three VaxStations and two Pericom Graphic
terminals. This gives a reasonable probability that a person attending a course may choose
to use the same type of terminal as they use in their own work. A class size is limited to
around twenty by the number of available terminals with two people sharing one keyboard.
This also seems a reasonable limit for the personal attention of a tutor and the size and
air conditioning of the office. In round numbers, if this facility were to be used for training courses for five days
each week and fifty weeks of the year then 2500 members of the CERN community, working on
the frontiers of modern technologies, could each receive 16 hours training per year. There may be a demand to enlarge this facility or to provide further units of a similar
size. The workstations have also been made available to experimental groups for private
demonstrations of their own analysis packages and for evaluation of workstations in
general. The processing power of the workstations may be exploited by appropriate
processing tasks overnight and as a low priority activity during the hands-on sessions. Video and computer based training have the advantage that the student may work at
his/her own pace, at any convenient time, without the presence of an expert or teacher. There is already some experience of the use of video training at CERN and in the
production of video training cassettes. CERN computer centre staff have received video
training in some basic VM operations and many find this an acceptable learning process.
Some of these videos were recorded specifically for HEP use [Footnote: B.White visiting
CERN from SLAC] and videos have been recorded at CERN for the safety training of the
members of each of the Lep experiments. " There should be a library of video and
computer based training courses at CERN. There should further be a public facility for a
single user, or a small group, to use this material. " Video and computer based training courses are available commercially for operating
systems, programming languages, data bases, word processing and so on. A selection of
subjects from a catalogue of 80 courses [Footnote: Computer Technology Group, Fox &
Elder, U.K. ] is tabulated. The videos runs for 20 to 45 minutes and the list price is
about 500 sfr (200 ukl). These courses are typically available without charge for one or two weeks and an
evaluation exercise may be mounted using video recorders and monitors already at CERN. An
opportunity is to evaluate a Unix course occurs when the CERN Apollo domain support moves
from Aegis to a the Unix version of the operating system. Video cassettes and computer based training material may be exchanged between CERN, the
regional centres and individual educational institutes of the HEP community. A multi-standard video recorder and monitor might complement the workstation clusters
of major experiments. This chapter covers, from a technology standpoint, the three major aspects of the
computational process, namely: the CPU, the data storage and the provision for data
transport (networking). It is not intended to cover all research activities, but to
investigate those which appear relevant to high energy physics needs in the 1990--1995
timescale. This section looks at the CPU technology that will be available in the next few years.
It appears that there will not be a clear division of labour as in the traditional data
processing centre. Rather, a functional co--operation between many different classes of
CPU will be the mesh on which physicists will perform their data processing. The following
sections broadly cut the CPU spectrum into 2 classes. The "high end" meaning
high performance, expensive CPUs in relatively small numbers, and "low end"
meaning cheap, plentiful machines. The reason for these broad categories stems from the
over use and now hazy distinction between PC--s, workstations, minis, mainframes and even
supercomputers. When considering a given class of CPU, it is worth remembering that
scalar, vector memory and I/O must all be in balance. They must be scaled to each other to
deliver the highest possible speed without system bottlenecks. Mainframes and supercomputers fall into this category. There seems to be no problem in
doubling the basic power of 1 CPU every 2--3 years for these machines. The emergent
gallium arsenide technology promises to make provision for smaller cycle times and higher
gate switching speeds, although commercial exploitation of this technology is only now
just beginning. This should eventually lead to higher basic CPU engines. For the future
(>10years), warm superconductors promise a further increase in switching speeds.
Manufacturers are also exploiting some parallelism in the current high end machines at the
operating system level. The IBM 3090 with six processors, the VAX 8840 with four and the
CRAY Y--MP with eight. It seems likely that this trend will continue as a way of providing
ever more powerful general purpose machines from tightly coupled multi processors. Vector
processing is now available on some IBM and CRAY machines to provide performance
enhancement at the user algorithm level. It is to be expected that both IBM and CRAY will
upgrade their machines, a 16 gigaflop CRAY--3 and a 64 processor CRAY--4 are rumoured,
whilst IBM have several parallel supercomputers under development. Despite disagreement
over architectures, there is no doubt that the design and manufacturing of parallel
systems is a growing industry. Much of this growth has been spurred by the demand for
cheaper supercomputers which has created the "crayette" or mini--supercomputer
market. The "preferred" architecture and methodology for exploiting such
architectures has not emerged to date. For HEP, exploitation of parallelism at the
algorithm level would not seem to be a general solution to the computing needs. Although
applicable, and highly beneficial in some cases, it would appear that running the same
algorithm many times in parallel would map better onto the bulk processing requirements. The evolution and development of VLSI technology, together with advances in computer
architecture have led in recent years to an upsurge in popularity of the RISC
architecture. As well as specialized architectures such as the INMOS transputer. The
potential here is for large numbers, of cheap, high performance, self contained processing
elements that can either co--operate in the processing of data, or operate independently
on data. The latter, massively parallel case, is of great interest to HEP. The emergent
Motorola 88000 microprocessor claims a performance equivalent to a single processor of a
VAX 8800 style machine. The workstation market has already adopted at its top end high
performance RISC machines capable of providing substantial CPU for the analysis of physics
events. For example, the APOLLO DN10000, recently announced, has a multiprocessor RISC
architecture that claims a performance of 50--100 times that of a VAX 11/780 style
machine. There are indications that probably all workstation manufacturers will have
machines of comparable power by 1991. The concept of the emulator farm for processing IBM
compatible object code images in parallel has been well accepted. Similarly, there is a
growing interest in using many workstations, networked as a parallel computing engine. The
growth in higher speed VLSI processors is expected to continue for the foreseeable future.
There will continue to emerge machines based on multiple processor architectures that will
attempt to exploit the parallel processing functionality with more or less success. Of
immediate interest to HEP during the next 5--10 years, will be the availability of such
processors to create specialized systems for parallel data analysis. This section looks at the current technology trends in mass storage devices and to
relate that to the future expectations of the HEP community. Firstly, it is appropriate to
review the current situation vis--a--vis mass storage usage in the CERN central
computer centre today. In particular, appendix A gives the current and expected figures
for the end of 1988. The data processing stream begins at the experiment. It is here raw data will be
captured. Processing of these data will produce the summary data that will subsequently be
reworked many times. Hence, decisions concerning technologies for mass storage begin at
the site of the experiment. We will consider in detail, however, the treatment of these
summary data. The model that is adopted here concerning future directions in mass storage
management will be the centralized one. That is to say a relatively small number of mass
storage data centres providing support for bulk mass storage media. However, any long term
mass subsystem must be able to provide connectivity to a wide variety of hosts (including
workstations) in a heterogeneous environment. The following points indicate some of the
key facilities demanded of centralized mass storage facilities:
Despite the strategic goals required to satisfy the mass storage access requirements in
the 1990s, one of the problems facing the HEP community is that of storage device density.
In this context, the following sections review briefly the current status and expectations
of the storage devices likely to play a role in the following 5 and in some cases 10
years. Evidence from research laboratories indicate that the magnetic storage technology still
has a long potential growth. Magnetic storage devices have, on average, increased density
at a rate of 25% per year. Higher density, higher bandwidth to the CPUs and higher
rotation speeds in order to improve seek/read/write time are essential in order to keep
pace with the increasing CPU bandwidth. Experimental magnetic disks have been made with a
track width of 20 millionths of an inch. This would potentially give a 3 1/2 inch diameter
disk a storage capacity of 10 gigabits, around 50 times the current density. The solid state disk is a device that appears to the CPU as a real disk but with a
greatly improved performance. Today, the unit cost of solid state disks is a factor of
10--100 greater than disk technology. However, recent solid state subsystems use the 1
Megabit chip and development labs have 4 an 16 Megabit chips available. In 5 years time,
with 16 Megabit chips at prices close to today's 1 Megabit devices the price gap between
solid state and magnetic disks may close considerably. The advantages of such devices in
terms of improved bandwidth, reduced latency and high reliability would make them
attractive, as magnetic media replacements in some applications, in the 5 to 10 year
timescale. Today the density of a cartridge tape unit as attached to the IBM systems at CERN
achieves capacity of approximately 0.2 Gigabytes. Given the data rates proposed by LEP,
and collider experiments, a major logistical problem in cartridge handling will result
unless a significant factor in storage density be achieved in the next 5 years. One might
already expect to see a factor 4 increase forthcoming in the short 1--2 year timescale.
From the technological point of view, the helical storage devices based on the video
cassette technology can already today achieve a storage capacity of roughly 2 Gigabytes.
An experiment taking roughly 35000 cartridges/year at today's densities might (hopefully)
expect this to reduce to 9000 in a short timescale (1--2 years) and down to 3000 in a
longer timescale (5+ years) There seems to be enough life left in the cartridge tape
technology to achieve a substantial increase in density. However this will not help the
record level access required to some datasets, as might be expected in a distributed
processing environment. This will lead to an increasing demand for tape staging and online
disk storage to aid this serial access technology. The promise of the optical storage media, namely, high density, random access, multiple
read/write capability and (potentially) high bandwidth has not been fulfilled in recent
years. However the write once/read many (WORM) technology has become very popular in the
text storage and retrieval business. Typically this offers today a storage density of
0.3--0.4 Gigabytes/side, data transfer speeds of 0.5--0.7 Megabytes/second and access
times of 100--500 milliseconds. The next generation of devices (available in the 1--2year
timescale) should offer 1 Gigabyte/side, transfer speeds of 1--1.5 Megabytes/second and
access times reduced to 50 milliseconds. If such performance and density becomes
cost--effective, this would appear to be a natural choice for the raw data taking which is
only written once. Subsequent processing of these raw data could profit from the random
access capabilities of the optical disk. By far the most interesting developments will in
the erasable optical disk units. There are today many companies working on this technology
and although some units will probably become available in the 1year timescale, it will
take some time before these units have the price/performance/reliability that will make
them interesting. The initial units will probably offer densities comparable with that of
today's cartridges (0.2 Gigabytes/side) and rather slow transfer speeds of 0.2--0.4
Megabytes/sec. However if in the 5 year timescale the technology matures to a density of
2--5 Gigabytes/side and transfer speeds in the 3--5 Megabytes/second (equivalent to the
maximum throughput of today's IBM channel) then they may well become the preferred storage
media for HEP. It is perhaps appropriate to inject some quantitative comments on the 'low end' of the
new technology coming. The digital audio tape, (DAT), like the compact disk, has also
found use in the PC and workstation market. A 120 minute DAT offers a capacity of 1
Gigabyte with a maximum access time of around 80 seconds. For the transportation and
subsequent analysis of data by PC--s this medium will probably become cost--effective in
the 1--5 year timescale for the individual, especially considering the falling cost
expected of the DAT units. This may open up new opportunities for data analysis. With the
increasing power of PC--s and workstations and the large volumes of data transportable by
DAT, "home analysis" could become very attractive especially considering that
running reconstruction programs on a PC should not be a problem in this timescale. Finally, in the research labs there are some developments that bode well for the future
5--10 year timescale. One such idea is the 'digital paper' concept. This is a flexible
optical recording medium with the sensitive surface coated onto a flexible polyester
substrate. This is today equivalent to the WORM media. The physical characteristics make
it suitable to be wound on to cartridges as does not have the rigid structure of today's
optical disk. Drives employ the standard 780--830nm lasers used for the rigid disk
technology. The projected capacity would be 600 Gigabyte to 1 terabyte, for a 2400 foot
tape, and the life of such a tape is quoted today at 15 years. Experimental tape drive
units, have apparently achieved 3 Megabyte/second transfer speeds and an average access
time of 28seconds to any byte on the tape. The packaging was a standard 12 inch tape reel.
Although it may be many years before commercial products based on this medium become
available, the data density, and transfer speeds seem to be encouraging for the future.
Such technologies are being driven by the industries that, like us, have the requirement
to store huge volumes of data for long periods (10years or more). From the above technology review, and bearing in mind the volumes of data to be taken
in the LEP, post--LEP and possibly LHC era a major problem of data management exists.
Since 1978, according to IBM, user demand for MIPS and associated storage have grown at an
exponential rate. From 1.5 Gigabyte/MIPS in 1978, through 4 Gigabyte/MIPS in 1984 to a
projected 12 Gigabyte/MIPS in 1990. With the volumes of data expected for the future
generations of experiments, automated handling of mass storage resources is a necessity.
Indeed, the era of distributed processing attempts to relieve the user of the knowledge of
specific devices and hardware configuration. We are starting to see the building blocks of
automated storage management such as the cartridge robot and the optical disk juke box,
with software tools such as the IBM Data Facility Storage Management Subsystem (DFSMS).
Such tools, aimed at a global system strategy have been slow to come, therefore we must be
careful to invest in system managed storage tools as they become available in order to
keep pace with the storage requirements of the HEP community. A totally heterogeneous
environment of system managed storage does not seem feasible until overall system
integration is tackled at a co--operative level between all manufacturers. The transport of data to and from the processing units and storage in a distributed
environment becomes one of the mechanisms that must have a bandwidth matched to the rate
of data consumption and production. The sections that follow review the technology both
for local "campus" style networking, and for wide area networking. are referred
to explicitly where relevant. Ethernet is the preferred local area medium today at CERN. Its wide acceptance in the
industry has made it attractive for connecting a large spectrum of heterogeneous
equipment. However, by 1995, Ethernet (10 Megabits/second) will be regarded as a
well--established, old--fashioned technology. Many Ethernet segments will be saturated by
distributed computing traffic. Many segments will be unreliable due to aging problems. The
IBM Token Ring will most likely be in a similar situation. The "normal" LAN for
serious distributed computing will by then be FDDI (100 Megabits/second). A typical
workstation or mainframe will be able to achieve a burst throughput at least 1
Megabyte/second on FDDI and FDDI costs will be comparable to today's Ethernet costs. On
the CERN site, the bottlenecks will be
1 gigabit/second LAN technology will be within sight but not yet available. As today's
unsolved problem is Ethernet--FDDI interconnection, the 1995 problem will be connection
between FDDI and a 1 gigabit/second backbone. On--site ISDN (at 64 kilobits/second) will
be available if we want it, but will be too slow to be interesting for scientific users.
It is unclear whether office automation users will still be interested by such low speeds
at relatively high cost. The number of network nodes will be so high (4000+ at CERN) that
configuration control and operational management will be a very major activity, comparable
to running today's computer centre. On--line data recording, if required, will not be able
to escape the mainframe bottlenecks. Hence, as today, transmission bandwidth will not be a
blocking factor. By 1995, 64 kilobits/second lines will be regarded as slow and their cost will not be
an issue. Basic rate ISDN (64 kilobits/second) will be in use to supplement traditional
public X.25 as a transport mechanism between small sites and for small sites to access
major sites. By 1995, 2 Megabits/second lines between major centres will be regarded as
normal and probably inadequate (indeed, NSFnet in the USA is already installing a mesh of
1.5 Megabits/second lines). At present international tariffs, HEP collectively would have
to pay at least 11 MSF/year for a set of ten 2 Megabits/second lines in Europe, not
including a transatlantic line. An approximate calculation shows that a comparable set of
lines in the USA, where prices are set by market forces, would cost only about 1.5
MSF/year. In the USA, there is reasonable hope that by 1995, trans--continental capacity
in the 1 gigabit/second range will be available to the research community in general, with
Federal funding, and presumably American HEP will be using it. In Europe, there is
reasonable hope that in some countries capacity in the 140 Megabits/second range will be
available to the research community on an experimental basis. The prospects of providing
such capacity on a multi--national basis at economic cost look poor at present,
essentially for political reasons. However, fibres with capacity of 565 Megabits/second or
above will be widely installed by 1995 and HEP should aim to exploit them on a
trans--continental basis. After the establishment of the single European market in 1992,
this may become economically realistic (since real costs are a small fraction, at most
10%, of current PTT tariffs). A political effort will be needed to achieve this. It is
completely unclear what will be the organization of international research networking in
Europe by 1995. It is quite unpredictable whether EARN will survive as a separate entity
and become a high--performance network. It is quite unpredictable whether the Eureka
COSINE project will come to fruition and produce a working international OSI network for
the research community, and if so whether the capacity of COSINE will approach the 1
gigabit/second, 140 Megabits/second, or even 2 Megabits/second level. It is therefore not
certain that HEP will still need to operate its own international networks by 1995, but it
seems highly probable. The only safe assumption is that we will be on our own as far as
high capacity links are concerned. By 1995, the OSI protocols over X.25 will be in general
use in Europe for routine applications (mail, file transfer, remote full--screen, remote
job entry). In America, the academic networks will have replaced TCP/IP by OSI, but not
over X.25. The HEP world will as a result be engaged in phasing out proprietary and ad hoc
protocols for routine purposes. Whether this will be a complete or partial process will
depend critically on whether HEP is still dominated by two proprietary operating systems
(in which case DECnet and SNA will remain of great importance), or whether Unix comes to
dominate. However, "modern" applications (particularly remote use of
high--performance workstations), and the need for bulk file transfer at very high speed,
will mean that proprietary and ad hoc protocols will also be in use. It is entirely
unclear that these will be based on X.25. Assuming a continued trend towards campus
networks, we can foresee two major performance bottlenecks in 1995 or earlier:
Thus one can assume that bandwidths significantly above 2 Megabits/second where
available, will be multiplexed between mainframe--mainframe channels and LAN--LAN
channels, rather than being shared by packet--switching. CERN is not alone in pushing the frontiers of technology for scientific data
processing. A recent report [Footnote: Datamation, April 15th 1987, page 60. ] on the NASA
Ames research center included their requirements for scientific computing in 1990:
Given the current technology status, this would seem optimistic for 1990, but
compatible with HEP needs for the 1990--1995--2000 time frame. In what follows the numbers are in Gigabytes. On VM a single density 3380 realistically
holds .53 Gigabytes. Looking at 1988 as a whole, the CPU capacity will have been doubled but the disk space
increased by less than 35%. The current total of 12.5 Gigabytes will be increased to 26.9 Gigabytes by the end of
1988. The distribution is : Use For the disk space for CRAY on VM the real figure of .53 Gigabytes per 3380 single
density equivalent has once again been used. By September 1988 there will be 25.5
Gigabytes VM/CRAY (on VM for CRAY) and 27 Gigabytes on the CRAY itself. In the table below
only the expected end 1988 situation is considered. The likely scenario is : Use Given the known increases as outlined above, by the end of 1988 the LEP experiments
could have approximately :
It is clear that these numbers fall far short of the 200 Gigabytes as called for by
MUSCLE for the end of 1989. This report has been prepared in response to a request from the Steering Committee for
the report "Computing at CERN in the 1990's" chaired by J.J. Thresher. To
prepare the report a subcommittee of six members was formed with as chairman Roy Billinge
who is the member of the editorial committee responsible for accelerator computing. The
composition of the subcommittee was as follows. Roy Billinge as chairman and providing the
input on MIS activities as chairman of SCAIP. George Shering as secretary and providing
the input on networking as a member of TELECOM. Lyn Evans as specialist on SPS matters,
and as a user of controls rather than as a builder. Claude Hauviller as liason with CAE,
in particular the mechanical design and engineering aspects. Fabien Perriollat as the
expert in controls and member of TEBOCO. John Poole as the expert on databases and to
bring input from LEP/CLIC on the requirements for simulations. The members were charged to
reflect their own views, those of their divisions, and those of the whole of CERN, not
only in their specialist fields but also over all the subjects discussed. The subcommittee
met six times, and this report summarises the discussions and decisions made. Networked workstations will become the generalised computing tools, with about 1000
general purpose office workstations and 100 special purpose workstations in the
accelerator divisions. Local and special purpose servers and networking support will
continue to be supplied from within the accelerator divisions, but the accelerator
divisions will increasingly rely on DD for central network support, for the main computing
facilities, for the accelerator database engine, and for backup facilities. For all this
they will expect as good support as the research divisions. It is agreed that CERN should
remain a multi-vendor site as a whole. The accelerator divisions, however, retain the
right to opt for a single line, either as a group or each division separately. Support for
open systems such as UNIX, and products that will run on a variety of support engines,
should be continued and even given preference. The accelerator divisions have a
substantial commitment to IBM compatible office workstations (PCs), mainly of European
manufacture. Network facilities and interconnections are being set up to capitalise on
their use. DD support for the supply and networking of this line should continue, and all
CERN-wide facilities made available on such networks. Coordination should be provided for
the choice of software to be used, and its relationship with other CERN software options.
Networking in the accelerator divisions is seen on three levels, the office LAN, the main
controls LAN, and regional LANs for equipment. A reduction in diversity of networking
technologies is desired. DD should provide strong central networking support, and should
handle all aspects concerning licences, distribution, and grouped ordering. DD should also
do the site network management, including allocation of names and IP numbers. A load of up
to 1000 packets per second between the accelerator LANs, and to the centre is envisaged.
About 100 home terminal connections will be required in the accelerator divisions. There
should be a single central database management system for all accelerators, preferably
running on the same computers. Over the period in question this should be ORACLE, which
will be increasingly relied on for operational requirements. Thus an ORACLE piquet, and a
piquet for its support engine, will be required. Accelerator design and modelling
techniques will move in part to workstation networks, but many large jobs will still
require the main computing facility. This could require between 0.2 and 1 CRAY XMP/48 year
equivalent spread over the five years 1990 to 1995. With the increasing power of
workstations on the one hand, and the better performance of networks on the other, the
accelerator control systems would be interested in replacing their local computation
facilities by on-line access to the central facility provided a suitable agreement on
availability could be achieved. CAE will continue to be important with Euclid on VAX being
the mainstay of the mechanical design, Autocad on the PC networks for mechanical drawing,
and PCAD on the PC networks for electrical design. It is foreseen that that over the
period the main priority for mechanical CAE will be the integration around a central
database of design, drawing, computations, manufacturing, and documentation packages.
PRIAM has become vital to the accelerator divisions. The support for embedded
microprocessor systems should increase, and DD should support an integrated software
workshop including tools for the specification, design, development and exploitation
phases. This should work in liason with ORACLE. The central documentation service should
be expanded to cope with technical and engineering documents, and should work in liason
with ORACLE. Increased manpower will be required in the accelerator divisions for
computing other than the control systems. In particular up to 15 people will be required
for local informatics and networking support, working in liason with the central support
services. Database application design will require another 15 to 20 people, which may
become less if one team could do all accelerator applications. The Accelerator divisions are involved in computing in two specialist areas, namely
Accelerator Design and Accelerator Controls, and in three more general areas of Computer
Aided Engineering, Database usage, and Management Information Systems. They were early
into computing with programs for lattice design, particle tracking, and field
calculations. With the advent of computer control of the accelerators, the accelerator
divisions had to build up a very considerable expertise in computer activities, including
things like busses, language design, networks, databases. This also overflowed into other
uses of the computers such as CAE and MIS so that the accelerator divisions often played a
leading role. With the increasing penetration of the computer into all areas of CERN's
activities, however, many of the accelerator control based activities became of general
application, especially as the experiments grew to almost accelerator size. Also there has
been considerable pressure on human resources in this area. Thus we expect to see a major
shift in the nineties to the accelerator divisions becoming users of more generally
supplied services, albeit very knowledgeable and discerning users. There is essential agreement on the trend towards networked workstations, together with
local servers, and increasing use of the central services over the network. It is agreed
that a "workstation on every desk" is a reasonable aim. In the Accelerator
divisions it is estimated that 1000 general purpose office workstations will be in use in
the Nineties. In addition there will be a number of special purpose workstations, Vax
Station or Apollo. Approximately, 30 will be used for mechanical design, 40 for controls
(20 in PS controls, 20 in SPS/LEP controls), 20 for accelerator design and modelling, and
10 for database design. An attempt is being made to minimise the use of special purpose
workstations, so a total of 100 seems reasonable. The accelerator divisions make extensive
use of a wide range of industrial software, and so tend to prefer the IBM compatible
office workstations for which there is the widest range of software and add on cards.
There are, of course, other reasons such as history, European element. The PS, SPS, and ST
have a policy on this matter, but in LEP both the IBM and Macintosh streams are present.
It was agreed that divisions should have a policy in this area, and an adequate mechanism
for implementing the policy. The advantages in concentrating on a single type of
workstation include better support from the limited manpower available, better network
facilities, easier communication, and a better spread of knowledge from other users. For
office workstation application software a two level approach is recommended. The
"high" level should be the best package for the specialist application (often,
unfortunately, the most expensive). The "low" level should be a popular package
running on the office workstation. Examples are found in many areas. In mechanical design
it could be Euclid and AutoCad, in electrical design Daisy and PCAD, in word processing
NORTEXT and a popular PC package, in databases ORACLE and Symphony or EXCEL, in software
engineering Teamwork and PCSA. At the popular level there is a danger of diversification
due to the individualism of divisions and groups. This might be controlled by
recommendations from the appropriate Technical Boards, by selective training and support,
and by providing interchange between the approved high and low end systems. Interchange is
made easier if the low end system is the same as the high end, eg. Euclid or Oracle on a
PC. This may not be the best answer, however, if the PC version is not a sufficiently
popular one. To avoid diversification the low level package must be popular and easy to
use. Management support is essential if a coherent advance in these areas is to be made. As seen from the figure, there are three levels of networking in the accelerator areas:
the office LAN; the Main Control LAN; and the Regional control LAN. For the office LAN,
the accelerator divisions are following the TELECOM recommendations and soon all offices
will be wired for Ethernet. It is thought that this will be adequate for the nineties. On
the traffic this may generate, on average only 20% of the users may be using their
workstation at any given time. However at peak periods, eg. morning, lunch, evening, up to
80% of the workstations may be in use for mail, looking at machine performance, or
checking up on equipment. Thus the network should be configured for almost maximum
utilisation at say one transaction per 10 seconds per workstation, maybe 10 packets per
transaction, say 1000 packets per second over the three networks with a substantial
fraction, say 50%, going to the central machines, mainly VM. The Main Control LAN is a
subject which has preoccupied control system developers since the first successful
introduction of the TITN network in the early 1970's. LEP have chosen the IBM Token Ring
as it has a number of advantages for control system use. Networking now has a very wide
application, however, especially within the large experiments. Much of the SPS/LEP
networking effort on telephones and TDM will be combined with the DD effort in these areas
in the future. The PS has only limited effort to devote to this subject. It was agreed
that the accelerator divisions would be happy that a single entity, eg. DD/CS, should
provide most of the LAN directions and support, with only support for local problems
provided within the divisions. DD should also provide support for controls protocols, eg.
TCP/IP moving later to OSI. This should include some measure of standardisation and
support for Remote Procedure Call protocols of which at present there are three, the
original one developed in PS plus two others, all different, one in DD and one in SPS.
Connectivity between the Main Control LAN and the office LAN should be provided. This can
take the form of MAC level bridges if the LANs are of the same type (Ethernet), or IP
gateways if the LANs are different but use the TCP/IP protocols. DD should handle all
matters concerning licenses and distribution of new versions, and provide grouped
ordering. DD should also manage the main network file servers, especially from the point
of view of backup. DD should provide the network management including names and IP
numbers. Some critical servers may be local, managed by the controls groups, but others
may be central and managed by DD. Increasingly it is believed that a regional control LAN
should replace the serial control highways more traditionally used. Connection between the
Regional LAN, through controlled gateways of course, should be possible to the central
database and microprocessor development facilities, and also to the office workstations
for monitoring and maintenance. Although it is agreed that Token Ring (a standard in its
own right), Uti-Net, Apollo Domain, and Apple Talk will be around for some time yet,
future decisions should move towards reducing LAN diversity where possible. Here again the
Technical Board recommendations should be given management support, in particular the
TELECOM recommendation for Ethernet. A number of home terminals will be required in the
accelerator divisions, in connection with the more or less well specified piquet system.
Although some engineers have expressed as desire to work from home, the mainline feeling
in the accelerator divisions is that the official allocation should be restricted to those
with on-line responsibilities. This will be about 30 per division, say 100 in total. The accelerator divisions are heavy users of databases of all types and the first main
application for Oracle was the LEP planning and installation. It is agreed that the
primary repository of accelerator information should be a large central relational
database. This is particularly important as the lifetimes of the accelerators and their
associated equipment are increasingly outliving a given person in a given post. The
lifetime of the database system should be at least ten years for reasonable stability.
After an intensive inquiry ORACLE was chosen and used for the LEP planning and
installation. It has grown until it now uses two big Vaxes, VXLDB1 and VXLDB2, running the
cluster version of ORACLE. There is a support effort for ORACLE in DD of about 5 persons.
Many users in the accelerator divisions are just now getting their applications running on
ORACLE. It is agreed that the support effort is such that only one database management
system can be handled at any given time. ORACLE Corporation seem to be keeping up well
with the technology, and indeed are the leaders in certain fields such as SQL
standardisation. It is therefore agreed that unless there is a major change in the market
place, ORACLE should be the database system through most of the nineties. For accelerator
data there was a consensus that there should be one single central database. This could be
quite a big load. This would imply some changes in DD operations. All major upgrades
should be made during accelerator shut-downs, ie. January-February. A 24 hour service with
on-line backup capability and an agreed maintenance service are required together with a
piquet service of Oracle and system expertise. The support required in the accelerator
divisions should only need to be applications expertise. It is important that ORACLE be
interfaced in a user friendly way to the office work station. Two approaches seem possible
here. One is to use distributed database access such as SQL*net which essentially make the
database available on the PC. The other is to use spread sheet programs on the PC and
provide easy automatic transfer between the spread sheet and the database. ORACLE are
working in both of these areas and the choice may be up to the user. Accelerator controls
have been reliant on on-line databases, often in the form of files in the control network.
It was agreed that the primary source of control data should also be the single central
accelerator database, along with all the other accelerator data. The PS are currently
moving their main control data base to ORACLE on VM. This raises two problems, speed of
access and availability. The availability problem should be solved in any case as
mentioned above. Several approaches to the on-line problem are being investigated. One is
to retain control system files and provide programs to swap data between ORACLE and the
files when changes are made. A second is to use a fast on-line database such as C-Tree and
again have programs to interchange data. A third is to use an on-line relational database
so that the links between the on-line and central database will be provided as part of the
manufacturer supported distributed database. It is not yet known which of these approaches
will be best, but the third will put more demands on the central database support group. A
solid and fast network connection is required, possibly based on an Oracle product. There
is a Remote Procedure Call interface to the Oracle VAX, but this is seen as a backup
solution. Priority for small accelerator jobs could be required. It is important that
Oracle be supported on a subset of agreed work stations with full networking, in
particular the workstations used for accelerator control. A tool to interface with
Teamwork would be required if teamwork becomes the standard SASD tool. DD should provide
the interface between the standard tools to get the best results. Currently there are
limitations on what can be held in and manipulated by Oracle. In the engineering database
it is required to store the "geometry" of objects, not just numbers and strings.
The whole concept has to be devoloped to handle this. At present work is going on to
interface Euclid to Oracle. There is a need for unification of accelerator data and this
will require concepts to handle 3D objects. At present dimensions are entered by hand. In
5 years or so we may have to replace Euclid, and it would be good if the eventual
replacement product could be Oracle compatible. Graphics capability is also needed for
applications in documentation, also for archiving and comparing waveforms to check if a
signal looks as it should. Oracle and Euclid should be on the same network, so that if,
say, we move a magnet using MAD we can see the new layout directly using Euclid. LEP and
SPS have a machine element database which is used by the machine, survey and installation
people. In the accelerator divisions, it is estimated that about 15 people will be
involved in database design, and up to 30 software engineers, expert in database
applications, will do applications programming. The number of designers may be reduced if
there is a strong central nucleus of database engineers giving support to all divisions.
On user feedback, Oracle has three user groups, the VM users, Accelerator controls, and
the application developers (mainly LEP so far). This arrangement seems to be working well.
When the PC users build up in numbers a separate group could be formed. User presentations
will become more important as Oracle expertise begins to spread. It would be very
convenient to have the same database for ADP and accelerators. Often in the design process
it is important to have access to lists of manufacturers, and in operation to know the
manufacturer, how many exist at CERN, contract numbers, etc. Although not so demanding as physics event analysis, accelerator design is increasingly
a heavy user of CPU cycles on the big central machines. Requirements include lattice
design, wakefield calculations, field calculations, particle tracking, 3-D structure
design. Linac design can be particularly demanding in these areas. At present the main
client is LEP with MAD, but as LEP moves into operation and more resources go into the
design of new machines, the computing requirements will grow. Some indication can be had
from the American labs. These range from 1000 hours per year of Cray time (SLAC), to one
Cray XMP/48 year equivalent (SSC). Much of the work is still done on IBMs, however, so the
load will be split. Accelerator design is becoming increasingly an interactive task
requiring good graphics and a fast turn round time. There is a demand for the use of
networked workstations, for example up to 20 Apollos for LEP, to meet this need. The
control systems are increasingly using modelling techniques in the on-line control of the
accelerators. At present these programs are run off-line, or on-line on number crunchers
attached to the control system. All this involves extra work and complication which would
be removed if the modelling could be done on-line on the central machines. This will
require improved and more solid networking connections, faster turn around for remotely
submitted jobs, and some declared availability. The availability need not be 100% as the
modelling need not be used in the primary control loop, but will be required for operation
mode changes, and optimisation studies. The availability maximum requirement might be
similar to that requested for the database machines. Some statement on the on-line
networked use of the IBM and CRAY are required from DD to enable the accelerator controls
to plan correctly. The accelerator divisions have a high concentration of engineers, and have contributed
heavily to the report on Computing for Engineering. Indeed more than half of the
representatives on the committee for Computing for Engineering were from the Accelerator
Divisions. Despite this, or perhaps because of this, some discussion and representation
from CAD is also included in this report on Accelerator Computing. This is to ensure that
the CAD influence on general computing requirements is not neglected. This is limited,
however, to mechanical CAD, databases, and the most common requirements of the control
system. To redo all the CAD requirements of the Accelerator Divisions would be to
duplicate the work of the Computing for Engineering committee. Not to mention this area
nor to consider its impact on workstations, networking, and the control system would
unbalance this report on Accelerator Computing. Therefore any mention of CAD requirements
in this chapter and elsewhere in this report should be considered as inclusive and not
exclusive. For a view of the requirements of CAD the specific report on Computing for
Engineering should be taken as the basis for requirements in the Accelerator Divisions.
The rest of this chapter is concerned with mechanical design. As mentioned above, the
subject of Computer Aided Mechanical Engineering is also treated in parts 3 and 4 of the
report on Computing for Engineering. CAD started at CERN in 1982 with the introduction of
EUCLID on a VAX780, and has expanded till in 1988 we have EUCLID-IS on a VAX8650 plus a
785 connected to about 30 dedicated work stations. This gives a total of 7 - 8 MIPS
computing power. In the near future we will introduce VAX stations in a Local Area Vax
Cluster (LAVC) with a 3500 file server connected to the cluster. DEC's new 6210 multi
processor number cruncher and database server will replace the 785, and 7 new VAX stations
will soon be put into service. Drawing started in 1984 with the introduction of AutoCad on
PCs. At present there are about 50 Autocad licenses at CERN, bought at prices ranging from
7000 francs down to 1000 francs, clearly a case for coordination. Of these 25 are in the
SPS, 3 in LEP, 3 in the PS, 1 in DD, and the rest in EP and EF. Work is in progress to
link Euclid and Autocad. Next year there will be a mini-Euclid on PCs which will be the
rough equivalent of Autocad, so Matra-Datavision themselves are not interested in links to
Autocad. More than 60% of the LEP drawings have been produced with Euclid. The layouts for
the whole machine are available on the database. In the SPS only 7 people, none engineers,
use Euclid. This could be because the SPS are doing less in-house design and using more
sub-contracts for design and development. The main use in the SPS, as in the PS, is for
maintenance type changes in layouts, and for detailed layouts of tight systems such as
target areas. There is only a little use of Euclid for CAM and links to the workshops,
mainly in ST. Within 5 years CAM could be vital to CERN, with the use of CAD in-house for
all design, and the diskettes sent to outside industry for manufacture. There will
probably never be more than 5 or 6 numerically controlled machine tools at CERN.
Communication is vital to be able to send designs around from system to system, even or
especially to different systems. There is an IGES standard which Matra-Datavision are
supporting which may help with this problem. Autocad was originally stand alone in use,
but now many of the AUTOCAD systems are linked over Ethernet using NOVELL in the PS and
SPS. The support required from DD could be quite different for the high and low level
systems. The high level system should be staffed and run by DD, whereas for the low level
system DD should organise licences, distribute software, provide courses, etc. There is
not yet a link between Euclid and the finite element packages used at CERN. Euclid is a
3-D package, however, and can calculate weights, inertias, and other geometrical
attributes. It is too early to have links with the finite element packages for stress
analysis, etc, but this would be very desirable in the future. This might mean moving the
finite element packages from the IBM to the VAX but this would put too big a load on the
VAXES. At present we use ANSYS which is an American package and CASTEM from the CEA in
France. There is a possibility of buying a CRAY vectorisation package for CASTEM. To
reduce the cost of using expensive packages for limited applications, it may be possible
to use them across the network in our sister institutes. For example a lot of the AA
design work was done using programs (TOSCA) on the Rutherford computers. This could add to
the wide area networking load. The current 7.5 MIPS of CAE VAX power will suffice in the
medium term, going up to perhaps 10 MIPS, but with increasingly powerful work stations at
the user end. A multi-purpose mid-range work station (PC with Autocad typically) will be
required per engineer, about 30 in total for mechanical engineers in the accelerator
divisions. A dedicated work station (Euclid type, probably VAXstation) will be required
per 1.5 to 2 designers, again about 30 in total. Software licensing will become important
as this could become an expensive item. Site licensing would be ideal, or perhaps a
license for N simultaneous users with no restriction on the number of potential users.
Personnel utilisation for mechanical CAD is at present 3.5 specific staff in DD plus
another 2 in supporting activities. Note that 2 of the 3.5 are leaving at the end of 1988.
Applications support in the divisions is about 4.5 full time equivalent, mostly on a part
time basis. In the future it may be necessary to use fellows and people on shorter term
contracts for CAD work. A new person can become productive after about six months
training, which is a reasonable investment for 1.5 to 2.5 years productive work. People
with a few years experience in CAD at CERN have no trouble finding jobs elsewhere. In the past the accelerator control groups have set up fairly powerful local computing
centres because their needs were new and specialised. These centres grew to handle jobs
such as accelerator modelling, controls program development, microprocessor software
development, controls documentation, MIS applications for the controls and operations
personnel. This will change, however, as CERN is setting up new central units to handle
things like MIS and CAE. The PS have two NORD-500s, one for modelling and one for general
applications. Also the PS have a local centre with two NORD-100s serving the secretariats.
Running these computer centres takes time and money and is something the PS feels could
better be done centrally. Originally the microprocessor work was done locally on the
NORD-500 but now this has been shifted to the PRIAM VAX. A lot of the PS Controls group
MIS work used to be done on the local computers but with the divisional emphasis on VM
this is shifting. The controls documentation is still on the NORD but this could move to a
CERN-wide scheme. For controls accelerator modelling, advantages are seen in doing this on
the same central machines as are used for the accelerator design, using as far as possible
the same programs. This will require, however, some decisions on the central machines'
availability for networked real time job submission. Thus the PS see the eventual
replacement of the local controls computing centre replaced by general central facilities.
On the other hand the SPS are happy with a local facility for accelerator modelling
(currently a NORD-500, possibly an Apollo DN-10000 in the future), as they find it easier
to get computing time and a good interactive service. Also it is easier to connect to the
control system. LEP would like to move to workstations for both accelerator design and
modelling. As mentioned in several other parts of this report, the new model of
accelerator computing is the use of personal workstations connected through a network to a
variety of resources. This applies to both the office workstations and the control system
workstations. The network resources take the form of servers. The control systems have
file servers for their data and programs. For office servers, both the SPS and PS use
NOVELL servers over Ethernet. These are used for PCAD, AUTOCAD, and general MIS
applications. LEP is likely to go in this direction as well. This concept is replacing the
traditional concept of the local computer centre run by the controls groups in the
accelerator divisions. The servers can be located close to the people providing the
support for that particular service, yet be available to anyone on the network, either in
the particular division, or even throughout CERN. Thus the NORDs will see their role
reduced to becoming servers on the network. In the case of the NORD-500s this will be to
provide a real-time on-line computational facility. As mentioned above, this function
could even be moved to the central machines using the network. The evolution of the
administrative support NORDs will be in the direction of becoming a server on the network
for the PS secretariats. A particular issue relating to local and central computing is
backup. It is agreed that backup facilities across the network to centrally provided and
run facilities is essential. This would apply to local computing centres, local servers,
and even office workstations. For the servers a daily incremental backup is required,
already done partially at the PS. An important item is the cross software support for the Motorola 68000 family. This
includes the RT kernel, at present RMS68K, and the cross compilers. at present C, Pascal,
Modula II, and FORTRAN. Also the linker/loaders, symbolic debugger, and executive
libraries. The aim should be to move to industrially supported products. The industrial
cross development facility for OS9, UNIBRIDGE, is urgently needed for large project work
in controls. The use of cross software must be carefully evaluated. It was generally
agreed that cross development was about an order of magnitude more difficult than native
development. Native development, however, requires provision for disks etc. at least in
the development phase. One of the advantages of OS9 is that it can easily allow the
development environment to be removed later. This was one of the driving forces for the
68000 in G64. An advantage of the cross software is that the central machine can provide
better facilities for project management. In any case the accelerator divisions will
require the PRIAM support for the 68000 and RMS68K to continue well into the nineties
until a replacement has been found, run in, and a reasonable changeover period has
elapsed. In accelerator control systems and equipment there is indeed a requirement for
embedded microprocessor systems, say on a single card, for which a cross development
environment is called for. However, much of the higher intelligence could be placed up a
level where native development systems are used. Anyway DD should support a native system,
for example native OS9 for the 68000 family. An Computer Aided Software Engineering (CASE)
toolkit should be supported by DD with full support for the accelerator divisions as well
as the research divisions. This should include tools for the specification, design,
development, and exploitation phases of big software projects. It should work in liason
with ORACLE. It should in particular cover the requirements of cross software projects on
the CERN standard microprocessors. The choice of tools is a difficult matter. Often they
are first chosen by user groups, eg. SASD, NOVELL, DV-DRAW. This is normal as the user is
usually the first to perceive a problem and see a solution in the literature. Once it
becomes established technology, however, it can be taken up and supported centrally. Who
then should drive the evolution? The user who still best perceives the requirements, or
the central support team who risk putting effort into things nobody wants? There was
consensus that the initiative should normally come from the users. Note that the
specification tools for databases are evolving on different but parallel paths with the
SASD tools for program specification. The hardware, work station and networking aspects of
the tools support should be set up, or at least specified, by DD. Accelerator controls are
big users of electronic CAD and CAE, but with no particularly difficult requirements. The
main uses will be for digital printed circuit board design. Services will be required for
programmable circuits such as EPROMS, PALs, FPALs. This should include support for the
compilers and provision of equipment for burning, equivalent or better than is provided at
present. The use of artificial intelligence and expert systems is expected to have a big
impact on accelerator controls and operation. At present a limited amount of manpower is
being put into exploratory work involving th KEE system. When this reaches the stage of
generalised application, considerable extra effort will be required to realise its
promise. At this time a central support for these new facilities will be required.
Controls requires a lot of documentation with the associated classification and printing.
Much of this is done at present using control system specific facilities. In the future
these should be provided on a general basis. Good quality (Laser) printing should be
available on the corridor level for all applications including control. This raises a
question of support for the corridor printers. An overall support operation could be quite
expensive in manpower, so continuing with an ad hoc combination of secretarial and
controls group support might be the only way to spread the effort. There is agreement that CERN as a whole should not be dependent on any single
manufacturer, and that competition is vital for good services and prices. Accelerator
controls have had a 15 year involvement with Norsk Data which has illustrated the good and
bad points of a single manufacturer policy. The SPS became very conscious of the bad
points, and after a period of in house designed computers, have settled for an "open
systems" approach. This involves the UNIX operating system and TCP/IP protocols, with
hardware available from a number of sources. To retain some of the benefits of a single
manufacturer policy the SPS has formed a close relationship with Apollo for workstations.
The PS have also felt the need to move to workstations and have chosen the DEC Vaxstation.
The PS have also decided to adopt the "open systems" approach with UNIX, and
will run ULTRIX (DEC's version of UNIX) on these computers. This provides a common system
across both controls groups, and will facilitate any future changes in hardware. DD
already has a number of support activities in the UNIX, TCP/IP, Apollo, and Vaxstation
areas, and clearly the Accelerator divisions expect this expertise and support to be
continued and strengthened in the future. For the general purpose office workstations the
accelerator divisions tend to prefer the IBM compatible PC types. Partly this is because
of the wider range of general purpose software available. Also, however, the accelerator
divisions are more stable and must take a longer term view on the re-usability of hardware
and software which favours the use of the most widely spread material. This may result in
a different choice from the research divisions which could prefer more state of the art
material. Nevertheless it is essential that all CERN services be accessible from a single
office workstation, so if there are "N-manufacturers" of office workstations at
CERN, many central services will have to be made available on all approved types of
workstation. There has been a feeling in the accelerator divisions that DD interprets its mission as
too much in favour of the research divisions and not enough in terms of support for the
laboratory as a whole. Hopefully this is changing with the establishment of the MIS unit
and Computing Support for Engineers. Also the Accelerator Control Groups have been a bit
"go it alone" in terms of computing technology, and again this is changing due
to pressure on manpower. Hopefully in the nineties all divisions will work together to
provide a range of agreed CERN-wide options. DD should see to all matters of purchasing,
licensing and registration for the approved range of products. An important role of DD
will be the networking and communication sitewide. It should be ensured that all equipment
can communicate well, and with the outside world. This could be a main guideline in the
selection of which material should be used. Technical support should be supplied by DD, in
particular for the central services such as the central interactive service, central batch
service, and central database facilities. However a local informatics support is also
needed, and in several examples from industry, and in the accelerator divisions
experience, a figure of 1% seems to be required, making about a dozen for the accelerator
divisions. This local support effort should work closely with the centrally provided DD
support activities. User Feedback can be provided on two levels. CERN-wide boards such as
SCAIP and TELECOM with the chairman outside DD seem to provide good guidance at the top
level. At the lower working level, boards coordinated by DD for special application areas
such as data acquisition, MIS, CSE, PRIAM, DEC user coordination groups, etc., are
appropriate. The resources allocated to computing within the Accelerator divisions can be divided
into three areas; Controls, Databases, and General Informatics. Controls are the dominant
consumer of resources at present, but Databases and General Informatics are increasingly
demanding. The accelerator control groups currently employ 90 persons for SPS/LEP and 44
persons for PS. Only a small fraction of this effort is directed to end user accelerator
control applications. These applications are, of course, essential and should obtain more
effort in the future. The small percentage of the total effort going to these end user
applications is due in part to the old SPS philosophy that end user applications should be
done by the end users themselves, ie machine physicists, equipment specialists, and
operations personnel. PS and LEP controls have, however, officially adopted the
professional applications programmer approach. The bulk of the controls effort has gone
into hardware, system software, and applications support structures. There is some hope
that rationalisation and the use of industrially supplied products will liberate more
effort for end user applications in the future. Database applications were pioneered by
LEP with ORACLE and are still limited in the PS and SPS divisions. As discussed in chapter
7, it is estimated that about 15 people will have to allocated within the accelerator
divisions to program applications if the database benefits are to be fully realised.
General Informatics support has grown up in a rather ad hoc way, partly within and partly
outside the controls groups. About 10 people can currently be identified as providing
informatics support in some way, but it is estimated that a more coherent effort of about
15 people will be required in the future to provide adequate local support, including
local network support as requested by the TELECOM recommendations. Provision of strong
central support in DD is essential if the above local support is to be most effective.
Also good central support is vital to achieve integration and economy with the rest of
CERN's computing effort. The accelerator divisions therefore support the requirements of
DD for effort in the MIS, CSE and Database areas. It is pointed out, however, that a
correct balance between central and local support must be achieved as no amount of central
effort can replace support close to the application. at CERN in the 1990s The main Working Group was composed as follows:
In the first Green Book "Computing in the LEP era", Computing for Engineering
was not recognised as a separate heading and the associated topics were barely touched on.
That five years later it should merit a chapter of its own, is a measure of the extent to
which the engineering community at CERN has come to use computers as tools in recent years
and the vastly increased sophistication of their needs. Indeed in several areas, such as
the use of formal software engineering techniques, they play a leading rôle in the
laboratory. As a corollary, strong requests have built up from the technical divisions for
increased support from DD Division. The working group which prepared this report was composed of the chairman of the former
Advisory Committee on Computing Support for Engineering (ACCSE) Each expert was left to prepare a section on his field as he saw fit. Most organised
formal subgroups which met several times to discuss the contents. Others simply prepared a
draft and circulated it to relevant people for comment. The drafts were then merged by the
convenor. In cases where they were significantly shortened, the original papers are
referred to in the bibliography. L. Casalegno, R. Dobinson, C. Eck, H. Muller, F. Perriollat (chairman), P. Van Der
Stok The community of microprocessor users at CERN and collaborating institutes is very
large, including both physicists and engineers, and it can be assumed that the impact of
this technology will continue to grow during the next decade. Some differences
exist between the engineer and physicist communities. The engineers are looking for
established industrial standards and products and they expect a system lifetime of 5 to 10
years. The physicists are more looking for frontier technology and for highest possible
real--time performance, to be able to cope with the expected data rate in the experiments
of the 90's. The lifetime of their systems is normally shorter. Nevertheless, it is
to be expected that the requirements of physicists and engineers can largely be met by the
same set of tools. The low end microprocessors (8 bits) should not be included in the central support.
Support must be concentrated on a limited number of microprocessor families; the M68K,
which is currently widely used, probably a reduced instruction set computer (RISC) family
(for example the M88K), and likely also the Transputer, if the current projects testing
its use will lead to a request for support to the User Committee. The Intel 386 family,
even if it will dominate the personal computer office automation market, will probably not
have a strong impact in real--time applications at CERN. This is due to the current very
strong engagement in the Motorola family and the fact that the market of VME boards with
the Intel 386 is not very active. This assumption must certainly be reviewed periodically.
At the board and bus level, activity must continue in the areas of evaluation,
consultancy and participation in standardisation committees. A key function, however, must
be to maintain close relations with industry to negotiate the best commercial and
technical support to CERN. As well as giving continued support for the VMEbus family, the
evolution of next--generation commercial bus systems should be monitored. For the difficult problem of system integration, which will continue to be the
responsibility of the user, some expertise and advice is expected from the central support
organisation. It is not, however, foreseen that the central support organisation will
provide much support for naked processors, nor will it be providing support for FASTBUS
and CAMAC equipment. For a native operating system like OS--9, which is currently very widely used at CERN
for the M68K family, the support should be mainly concentrated on logistics: licences,
distribution of software and documentation and the organisation of user fora. The future
evolution of this type of system, e.g. bridges to UNIX and VMS, will be followed. The next years will see important new industrial products emerging in the area of
host--target development systems. A high quality and powerful environment must be provided
for this class of systems. The chosen environment must be available for DEC's VAX--VMS,
for UNIX and probably also for workstations using the IBM/Microsoft OS2 system. It must be
well integrated with the selected Computer Aided Software Engineering (CASE) tools. This
environment must be able, for new applications, to replace the currently supported Real
Time kernel RMS68K in the near future. Important criteria for selection are the quality of
debugging facilities for programs and systems, and good integration of communication. The
DD central support group should continue to make this development environment available on
a computer (or a cluster of computers) used as a server by user workstations as well as by
users who only have access to a simple terminal. The existing system, a VAX 8530 running
ULTRIX, will have to be regularly adapted to growing demands on its capacity. It is
assumed that over the next three years most of the additional computing power required for
this development style will be provided by workstations, bought by the users and networked
to the central system. It will thus probably be sufficient during this time to expand only
the disk capacity of this system but a major upgrade still has to be foreseen in the first
half of the 90's. It should be noted that the PRIAM VAX supports a number of general
services which are not connected to microprocessor support and it will continue to do so
in the future. The programming languages will continue to be C and FORTRAN, with perhaps PASCAL in
addition. Modula 2 is not much favoured. There is already pressure from the technical
divisions for ADA support in the 90's and it can be assumed that the physics community
will follow this lead. Interpreted languages, like NODAL and PILS, are well accepted as
testing tools in both engineering and physics applications. Based on lower level tools
provided by the central microprocessor support, these languages have been developed in
CERN groups working directly for end--user application support. It is assumed that this
remains so in the future. It is clear that the implications of attempting to offer support
for a RISC processor and/or Transputers must be evaluated. It seems to be unavoidable
that, during a trial phase, a number of incompatible processor architectures will be used
at CERN. Nevertheless, based on the experience gained in this field, this number should be
reduced as soon as possible to the smallest value compatible with the needs of the user
community, in order to be able to give a reasonable level of support at all. A centralised CERN--wide support for microprocessors is strongly recommended in order
to reduce duplication and to increase the quality of the service provided. The effort must
be concentrated on integration of well established industrial software components. No more
in--house development of basic components seems to be necessary. Expert advice to the user
remains an important aspect. Unless, however, the central service is properly funded and
staffed, it risks to be ineffective and alternative ad hoc solutions will inevitably
appear. It is recognised that the microprocessor user community at CERN will only be able to
cope with the continuous evolution in this field, if sufficient training possibilities for
all levels of usage of the supported tools are offered. It will be the task of the central
support group, in collaboration with CERN's academic and technical training services, to
provide the necessary courses for this. C. Hauviller (Chairman) This part is intended to be a summary of a series of papers on the future for Computer
Aided Mechanical Engineering published during the last two years Computer Aided Design (CAD) has been considered for long as a stand--alone subject and
this still remains often the case at CERN, but the general evolution in firms of any size
is towards as complete an integration as possible of the very diverse mechanical and
electrical engineering activities. To extend this notion of integration to CERN does not
seem unrealistic, even if it must be adapted to the organisation's specific activities and
environment. For this reason, the present section, although mainly dealing with CAD, also
emphasises the larger notion of Computer Aided Engineering (CAE). After quite a slow start in computing for mechanical engineering, software adapted to
accelerator and detector technology is now widely used at CERN, especially for the present
very large projects. In the mid 70's, the first finite element method (FEM) packages were
installed on the CDC 7600. Ten years later, packages provided by software firms are
resident on the central computers, together with smaller ones installed on PC's. This
subject of Structural Analysis is dealt with in Part The next step in the introduction of computer intelligence in mechanical engineering
has been numerically controlled machining. This has, however, been limited to a small
number of machine tools by the fact that only a very small amount of equipment is and
probably will be actually produced in--house. The major breakthrough has been CAD. The decision to install a CAD system in the
technical divisions was made in 1981 and one year later, after a detailed technical
evaluation, a true 3--D solid modelling system, EUCLID from Matra Datavision (MDTV), was
selected as an interdivisional project. The choice of a 3--D package oriented towards
design instead of a 2--D (drawing) one was difficult. The market trend was still towards
the simplest packages and solid modelling software was in its infancy. It was considered,
however, that the potential usefulness of solid modelling was much greater, in particular
the assembly capabilities allowing the integration of various parts of accelerators and
detectors. EUCLID was installed on a VAX 780 with workstations in the form of simple
storage tube graphic terminals. Six years later, the new user--friendly version, EUCLID--IS, is running on more than
thirty dedicated workstations connected to two clustered VAXs, a 8650 and a 785. More than
60'000 geometrical entities are stored in the site--wide database. The computing power has
been multiplied by 8 since 1982. Besides this powerful design system, a simpler drawing package, the very widespread
commercial product AUTOCAD, has been introduced recently. It runs on IBM--PC compatibles,
usually in a stand--alone configuration. About 35 licences of this software are in use at
CERN. Both packages are complementary and have their own advantages: EUCLID, built around a
central infrastructure, is more powerful but AUTOCAD is cheaper and easier to learn. The
former is adapted to large projects which have to be co--ordinated and require a major
database whereas the latter is preferred by individuals working on little stand--alone
projects. Moreover, both are in use in the High Energy Physics community. Mechanics is the main application of these packages but EUCLID is also used for
infrastructure services such as civil engineering, electrical engineering, survey (in
addition to the ESPACE and LILIAN packages), etc.. Communication is a must in large engineering projects like accelerators or detectors.
It is, therefore, unthinkable that CAD should live in autarky. The general industrial
trend is towards as complete as possible an integration of the various engineering tasks.
The aim is to gather around a unique database, under the term Computer Aided Engineering
(CAE), diverse activities such as design, computations, drawings, schematics, simulation,
manufacturing, maintenance, etc.. Only design and drawing are considered here. All the administrative tasks such as
budget, planning, etc. are and will be done using tools like ORACLE, available on a
CERN--wide basis accessible through a local or general network. The population involved in mechanical engineering is presently about fifty engineers
and more than hundred full--time designers It can be assumed that this population will remain constant or decrease slightly during
the next decade. However, many of them are not yet very familiar with computers and so it
can be expected that the CAE user population will grow. In addition, subcontracting will
take more importance and this stresses the need for good means of computer communications
towards the outside of CERN. After an initial period of investment, only a maintenance budget was awarded to the
EUCLID project during two years. The consequence of that situation was a lack of
continuous hardware upgrading. This upgrading has only been re--started during the last
year and the expansion of the system (number of users) will only resume by the end of
1988. The total number of dedicated workstations is likely to be about 50 by 1990. This upgrade is planned in four steps:
The EUCLID software has been continuously upgraded toward a more user--friendly
product. A very powerful 2--D package, available at the beginning of 1989, will provide
functionality equivalent to AUTOCAD. Some 3D features have appeared in the latest versions of AUTOCAD. In addition, the
power of the IBM--PC compatible platforms is increasing. Budget constraints presently
limit the number of users of EUCLID and so AUTOCAD use at CERN has doubled during the last
twelve months. Since AUTOCAD is considered to be a good introduction to CAD and a standard
in collaborating institutes, this trend will probably continue for a short time. However,
the difference in investment between the two products decreases rapidly and will become
negligible in less than two years. This will lead to an equilibrium between the two
packages, the choice depending of the type of work, its aim and the environment of the
user. To support more than two packages (EUCLID and AUTOCAD) will be too expensive (direct
and indirect costs) and inefficient. Introduction of new programs must be avoided except
if it can clearly be demonstrated that their features are much better than the existing
ones. However the official choice should be re--examined periodically in terms of
technique and finance, especially if a new large project starts. All the numerical control applications have been integrated in the CAD environment
through EUCLID under the same user interface. The numerical control language available is
APT. The present capabilities satisfy all the needs met until now and it is recommended to
proceed in this direction. Future Computer Aided Manufacturing (CAM) subcontracting will
also be based on the same principle. Integration should be the main priority to avoid duplication of work and stored data,
especially for large projects. The 3--dimensional geometric description provided by EUCLID
already allows partial integration but a complete object definition encompasses more than
geometric data. Technical (specifications, tolerances,...) and administrative (drawing
number, supplier,...) data should be added. The open structure of a Database Management
System (DBMS) like ORACLE is the obvious core for such an integration but standard
databases must evolve to handle correctly the combination of geometric, text and numerical
data found in engineering. The ACCSE report F. Bertinelli, M. Bona, D. Jacobs, A. Lefrancois, R. Maleyran, A. Poncet, M.J. Price,
D. Wiskott(chairman). During the last 15 years a number of engineers and designers have gained access to
programs that permitted them to analyse structures for their mechanical and thermal
properties. These structural analysis (S.A.) programs were obtained either through
purchase or with the help of other institutes. They perform calculations on models that
approximate the structures with an assembly (mesh) of finite elements. In the beginning,
pre--processing (mesh generation) and post--processing (result evaluation) was done in a
more or less ad--hoc manner. The matter was an affair of specialists who wanted to solve
their specific problems. These programs run on the CERN mainframes, originally CDC
computers. Some of them (e.g. SAP--V, DOT, BOSOR) were later also transferred to the IBM
under VM or to VAX under VMS (often without the interactive facilities that meanwhile had
grown around them). Beside the older programs, three products are used at present: ANSYS, sold by Swanson
Analysis Systems Inc.(USA), CASTEM, designed by CEA--Saclay (F), and CASTOR, written by
CETIM (F). For ANSYS, CERN profits from very advantageous (educational) commercial
conditions. With the CEA, CERN has a development contract for CASTEM (and therefore has
also access to the source code). ANSYS and CASTEM run on IBM under VM, and a reduced
version of ANSYS can also run on PCs under MS--DOS. CASTOR runs on VAX and PCs. For the
first two programs updates are regularly received from the suppliers of the code. Entirely
new code for CASTEM ("CASTEM 2000") has been received but only partly been
installed yet. CASTOR, limited to 2D problems, has been acquired as a simple tool to serve
the needs of users in drawing offices. The number of users of these programs, in particular CASTEM and ANSYS, is increasing
slowly but steadily. This increase corresponds on the one hand to the growing complexity
of the engineering tasks, where performance is often pushed to the technical limits, on
the other to the desire to have more insight into the behaviour of structures conceived in
day--to--day work, to reduce the necessity for prototyping and to optimise the final
design. Increased acceptance by the users due to improved program performance plays an
essential rôle as well. The number of users, excluding occasional ones, is around 30 engineers, designers and
applied physicists. Fellows and visitors (who carry a substantial part of the work load)
are counted in these figures. Support for the user population is given either by experienced users or through them
(mainly on an ad--hoc basis and for special informatics problems) by the DD user support
(DD--US) group. A number of courses have been given on the theoretical basis of structural
analysis and on the use of ANSYS and CASTEM. Considering a given user, work with S.A. programs is not a continuous activity. Very
often, however, solutions of some problems require substantial resources in computing
power and disk space for a limited time. It is not always easy to find these resources
readily. Extrapolating from the present population and assuming improvements in all kinds of
user support, the number of regular users, during the next five years, probably will not
go beyond 45. It is probable, however, that their access to S.A. programs will become more
intensive and more demanding on resources, as they will have to face the very involved
design problems of future accelerators and detectors. Simultaneously, with better user
interfaces and improved support, the community of occasional users will grow, but this
activity is difficult to quantify. The working group intends to study the technical features of a range of programs. It is
agreed that, for the selection of general purpose programs to be maintained or introduced
and supported at CERN, particular emphasis, apart from powerful algorithms and needs for
resources, should be put on user friendliness and interactiveness. Versions with identical
user interface (syntax, man--machine dialogue) should be available for different levels of
problem solving power (e.g. running on PCs, workstations and large mainframes). Even for
problems that demand mainframe computing power, pre-- and post--processing should as far
as possible be shifted to workstations or PCs. Structural analysis in the above defined sense should not be seen as an isolated
domain. Integration with other aspects of CAE is becoming increasingly desirable. Magnetic
and electrostatic fields as well as thermal effects have repercussions on the behaviour of
mechanical structures. Modern CAD systems offer a natural method to describe the geometrical properties of a
structure to be analysed. With the tendency towards distributed computing and with the increasing computing power
of independent units, it can be estimated that a large fraction of the problems can be
solved on workstations or even on personal computers. Fast network connections should
provide easy transfer of the program units needed from a server station. For extended
problems the same network should provide access to the large mainframes. It is expected that the availability of large computers with vector processing
facilities opens up new possibilities for S.A. calculations. Larger problem sizes will
become accessible and the present day problems, especially where iterations are involved,
will be solved in much less real time. The present situation suffers from the fact that on the informatics side no explicit
user support is available. On average each of the estimated 45 regular users will use some 40 hours of
168--equivalent computing time per year (= 1800 h/y in total) and will need a permanent
storage allocation on disk of about 15 Mbytes (= 675 Mbytes in total). As hardware is assumed to be provided by the user divisions and the computer centre,
and expenses for purchase and maintenance of programs and documentation are covered by
Part D. Cornuet, D. Dell'Orco, K. Freudenreich, R. Fruhwirth, H. Henke, A. Ijspeert, R.
Perin, M.J. Price, T. Tortschanoff, J. Tuckmantel, H.--H. Umstätter (chairman), J.
Vlogaert A fuller version of this part is to be found in The most widely used packages are POISCR (a CERN version of POISSON) and MAGNET. These
employ two different methods: MAGNET the older finite difference method with square
meshes, POISSON the finite element method. Unlike MAGNET, POISSON admits cylindrical
co--ordinates, plots fields, computes magnetic forces and is also useful for high voltage
engineering, electrostatic separators, septum magnets and wire chambers since potentials
can be specified on mesh--points. Its main advantage is inherent in the finite element
method with triangular meshes of variable size. Its main drawback is the lack of a
graphics preprocessor. This makes it necessary to enter the mesh by hand as a list. MAGNET
is very precise, easy to use due to good documentation and, despite its limitations, is
still in use because data input takes less time than for POISSON. MAGNET has no graphics
output at all. The 3D finite difference program PROFI is available at CERN. Although difficult to use,
it allowed precise 3D calculations whereas the earlier GFUN was more flexible but less
precise. CERN's versions of both packages are old and lack graphical input. There is no commercial offering in this domain. We list here the main packages in use
at CERN:
In view of the lack of commercial products, central informatics support is especially
important for this class of users, mostly for installation, updating, help with graphics
and workstation selection. The two main problems of the old software available at CERN; lack of pre-- and
post--processing and lack of good 3D programs, are being solved by the introduction of
TOSCA and ANSYS. TOSCA from the Vector Field (VF) company, Oxford was the most popular choice in the
ACCSE user inquiry and is now installed on VM--CMS. It is well known among accelerator
magnet builders and much experience has been accumulated worldwide and at CERN. In
addition, it has modern graphics pre-- and post--processors, can solve both 2D and 3D
problems and can make use of the CRAY for the main calculations. In the future it may be
interesting to acquire additional software packages (e.g. CARMEN or PE2D) for eddy current
prediction in pulsed magnets. These use the same pre--processor as TOSCA. ANSYS (see Part The 50 or more staff active in the field require central informatics support at a level
which they estimate as involving two persons. G. Baribaud (chairman), A. Fowler, B.I. Hallgren, E. Heijne, P. Jarron, K.D.
Lohmann, C. Schieblich, E. Vossenberg, R. Zurbuchen. In the area of accelerator development and design there is already a long tradition for
the use of CAE. In the field of analog electronics, on the other hand, CERN has to make a
considerable effort if it wants to keep pace with the introduction of CAE/CAD--based
methods already in use in industry. It is felt that a global introduction of these
methods, applied to the total engineering cycle from conception and design to acceptance
testing and documentation, would necessitate a reorganisation of all related activities,
and require investments and the hiring of specialists. Time did not allow us to develop
alternative well--documented scenarios for the transition of CERN's analog electronic
activities into the age of computer aided engineering but this is clearly a high--priority
task for a future working group composed of representative users and support personnel. It
is anticipated that a major trend will consist of a drastic shift from conventional analog
electronics towards Application Specific Integrated Circuits (ASICs). In view of this
situation, and the very limited time spent in assessing future requirements, it is
difficult to make forecasts beyond the next 3--4 years. At CERN, Analog CAE/CAD covers a wide range of application areas Some important application areas, such as for instance signal processing, have not been
covered by the working group because of lack of time and expertise. Four main areas of
CAE/CAD, each characterised by specific requirements, have been identified, and are listed
in no particular order. The complete cycle for the design and manufacturing of ICs depends entirely on computer
aided methods. The manufacturer normally provides the IC designer with an integrated set
of tools, tuned and tailored to the fast--evolving technology, together with a
process--specific design database. The designer needs a workstation--based, high quality
working environment. During the course of design optimisation, a great number of
simulation and layout iterations have to be performed. Hence adequate computing power for
simulation is of prime importance and access to mainframe--based simulation is very
essential. The simulator used to this end must guarantee to keep pace with the fast
evolution of IC technology and provide for up--to--date device models. For layout work,
high resolution colour graphic displays and the use of a high quality colour plotting
facility is also a requirement. Another area of CAE--based activity related to
microelectronics is the characterisation of a large number of semiconductor devices, both
on wafers and in packaged form. This requires a laboratory equipped with suitable hardware
and software. Furthermore, assembly and test equipment in a clean room environment must be
available. For analog electrical and electronic circuits made of discrete components assembled and
mounted on printed circuit boards, breadboard prototypes are still widely used during the
design phase. Analog simulation can favourably replace the breadboard, provided the
designer can find his component models in a component library. At present, no adequate
component library is available for that purpose at CERN. The usefulness of a library
depends strongly on the availability of component models and their quality. The
procurement of semiconductor component parameters and other coordination tasks requires a
highly qualified and experienced engineer. For efficient discrete component analog circuit
design, circuit simulation must be complemented by schematic entry and printed circuit
layout CAD tools. Analog circuit layout is often very critical and requires a close
collaboration between the designer and the layout specialist. Their proximity and the
integration of the design and layout tools is therefore of prime importance. In this context, a system can include elements of very different nature and involve
several disciplines; electrical, mechanical, thermal, etc.. Discrete component circuits,
sensors, actuators, high power switching devices, superconductors, plasmas, high power
conversion and distribution equipment, etc., can be parts of the system. The use of a
physical prototype is normally impossible or impractical and analog simulation must be
used as the means for design optimisation. The emphasis for the analog simulator must
clearly lie on flexible and versatile modelling capabilities, allowing behavioural and
functional descriptions, along with the possibility to optimise for specific design
criteria. The fact that a large part of system design at CERN concerns equipment for
accelerators, calls for a long lifetime and good stability of these design tools. RF and Microwave design differs from more conventional circuit design in several ways.
The influence of component interconnections cannot be dissociated from the overall
behaviour of the circuit, and has to be modelled accurately by using measuring equipment
coupled with the simulation tools. Tools for circuit layout and mask generation are also
needed, and should be linked to a mask cutter. The use of high frequency component models,
and their strong dependance on circuit layout, means that special device libraries are
required. Finally, frequency considerations impose special circuit descriptions in terms
of wave parameters. This leads to highly specialised and integrated RF and microwave
CAE/CAD tools, running on workstations in conjunction with special laboratory equipment. [Ref.] P. Bähler, B. Flockhart, L. van Koningsveld, G. Leroy, M. Letheren, A. Marchioro,
L. Pregernig (chairman) This part will ( At CERN, engineers and technicians design electronic circuits covering a wide range of
applications, complexities, and technologies. Fairly detailed knowledge exists about the
different categories of design work at CERN. In 1987 designers ordered several hundred new
board designs from CERN's printed circuit board (PCB) workshop and 30% of them were
"high density" jobs. The high density jobs may account for about 70% of the
total time spent on designing electronic circuits. It is expected that in--house design
efforts will continue at more or less their present level. Designers are lacking skills in
new design methods and tools to establish and to verify system descriptions, such as
functional descriptions, which can be used for subcontracting. CERN employs about 450 people in staff categories 204 and 304. Maybe one third of them
are hardware designers. Often, they work at the group or section level of the CERN
hierarchy and few official communication channels exist for them. Some designers have
access to modern computer aided engineering (CAE) tools which CERN has purchased in the
past. But so far, less than 10 people have received formal training to the level where
they are proficient in all the capabilities offered by the tools classified as Class 2 in [Ref.]. At present, CERN's layout experts are reasonably well equipped with board layout tools
and they meet regularly to coordinate their efforts. However most of our system and
circuit designers lack (i) modern design tools, (ii) skills in modern design methods, and
(iii) the infrastructure to help them to apply such tools and methods. It will require a
co--ordinated effort from designers and management to keep up with the evolution of
electronics. Recommendations about CAE/CAD tools for the 1990s also need to consider trends in the
electronic industry. The next section describes those that are likely to influence
electronic design at CERN. Ten years is a very long time in electronics and therefore it is difficult to predict
accurately how design methods and tools will change beyond the next 3--4 years. However,
one can identify 3 major trends in industry that are likely to influence electronic design
at CERN in the near future. Firstly, more designers will use Application Specific Integrated Circuits (ASICs), like
complex programmable logic chips, gate arrays, or custom chips. Predictions are that 80%
of all designs will contain at least 1 ASIC by the year 2000. Secondly, more designers will apply "Top Down" design methods, i.e. they will
use hardware description languages to describe architectures and functions of complete
systems before starting any detailed design work. For instance, VHDL
(Very--High--Speed--Integrated--Circuits Hardware Description Language) has become a
standard (IEEE--STD--1076--1987). Unlike text specifications, such functional
specifications are "executable". One can verify them by simulation. Furthermore,
major research efforts aim at developing programs which convert functional specifications
(descriptions) into detailed circuit diagrams and/or layout (logic synthesis and hardware
synthesis). Today, some companies already use synthesis tools successfully. Major CAE
vendors are integrating new classes of tools (e.g. hardware description languages or
silicon compilers) with their existing products. Thirdly, components housed in new types of package, e.g. surface mount devices, etc.,
will become more widely used. This trend, together with increasing device speeds, will
make board layout a more difficult task. The recommendations in this section suggest how CERN can cover the needs of its
designers, keeping up with the evolution of electronics. The recommendations consist of 2
parts, a resource and a management issues section. The working group mainly focussed on
design methods and did not consider in detail matters such as computer aided testing,
computer aided manufacturing, networking and the present generation of programmable logic
tools. One way to estimate the number of tools required is to assume that:
Under these assumptions CERN should gradually introduce approximately 20 additional
"Class 2" workstations (13 already exist). Ideally, "Class 1" and
"Class 2" tools should be for personal use, i.e. designers need access to them
without sign--up procedures. It is also assumed that improved design methods will help to reduce (or to maintain at
the present level) the workload for Class 3 tools. For instance, formal design methods
(behavioral descriptions, top--down approach) are a means to avoid redesigns due to poor
initial system specification. Furthermore, simulation helps to eliminate design errors and
therefore reduces the number of iterations in hardware prototyping. Finally, shifting
design capture tasks from Class 3 tools to other classes would effectively increase the
resources available for layout tasks. Informatics
and operational support involves keeping installed systems running, installing new
systems, upgrading systems, developing and maintaining interfaces, tracking down bugs,
running an electronic bulletin board (for information exchange), etc.. Library
incompatibilities cause major problems porting jobs from one system to another. It
requires considerable effort and ongoing support to solve such problems. One should
therefore avoid interfacing different systems, whenever it is possible. If commercial
interface standards become available (e.g. EDIF), the recommendation is to adopt them. CAE
vendors (and third--party vendors) offer libraries but they are incomplete. Therefore, a
team of experts, e.g. cluster leaders and library coordinators, should assume
responsibility for CERN's component libraries, (i) establishing strict rules for component
creation and (ii) monitoring their application. Applications support means to assist designers in using CAE tools and to ensure that
they follow guidelines (e.g. for component creation etc.) in applying the tools. Modern design tools are complex and a designer who works in isolation is less likely to
use them as efficiently as a designer working in a team. Such a structure is modular and therefore expandable. For instance, after an initial
pilot project, one can gradually increase the number of design clusters according to the
design requirements. Furthermore, when new tools become available, one could test them on
one cluster, before approving them for general use. If the pilot design cluster is successful, one could expand according to requirements
(e.g. one design cluster per year). The capital requirements would be some 500 kSFr. for
the first phase and about 750 kSFr. for a subsequent six--workstation cluster (at November
1988 prices). This suggestion assumes that groups will purchase Class 1 tools on their
individual budgets. The recommended approach aims at systematically introducing modern design tools and
methods at CERN. Implementing it will generate benefits for CERN, such as:
J. Schinzel (chairman) The number of people involved in database application development for engineering is
fairly extensive and many of these were very much occupied with the LEP Injection Test at
the time when this part was being prepared. Therefore rather than forming a committee, the
draft documents were circulated to the people known to be involved or interested in the
engineering application software using databases. The many comments were gratefully
received. During the 1980s the usefulness of Database Management Systems (DBMS) has been
generally accepted by engineers. The rapidly expanding user community of the Relational
Database Management System (RDBMS) ORACLE both on the VAX/VMS and IBM VM systems and the
growing number of data management applications suggests that database technology will be
increasingly exploited in the 1990's. One can assume that during the next decade ORACLE
will continue to be offered to engineers at CERN, since there is at present no
significantly better product on the RDBMS market. However, in the database field,
technology is evolving so rapidly that it is dangerous to predict too far ahead. Increased
development support will be required for all fields where data must be managed, whether
engineering, accelerator control, physics experiments or Management Information Services
(MIS). CERN's investment in application development and data entry is growing rapidly. The
full benefit of this investment will only be realised if there is CERN--wide coordination
and management of all data and software. More emphasis will be placed on reliability and
ease--of--use. Compromises may be necessary between stability and the introduction of new
tools. Workstations are already being used for database application exploitation. As the
number of remote processors accessing databases expands, so will the need for more
sophisticated network management tools and support. Around--the--clock access to data will
have to be considered, as essential services depend more and more on database technology. ORACLE was installed on a dedicated VAX in 1982 for the management of LEP construction
data. Applications include project planning, installation logistics tools and systems for
managing inventories, cables, drawings, documents, magnet data, etc.. Many facilities
share common data such as lists of CERN personnel, machine elements and locations. The
control database presently in development will access much of the data collected for
equipment installation and testing. Approximately 400 accounts have been requested on the
LEP DB (LEP database) VAX, including 'public' multi--user accounts. About 40 LEP and SPS
staff are active in developing applications of which approximately 30 are in production.
The production database contains well over half a gigabyte of data. ORACLE was introduced on the public VMS service in 1984 and on the IBM VM system for
general CERN use in Autumn 1985. At present there are about 200 accounts, approximately 40
of which are in constant use. The applications treat both administrative and technical
data. The Database Section in DD--SW Group is responsible for database administration,
software distribution, installation and support to application developers in VMS, VM and
workstation environments. Support includes advising users on technical ORACLE problems and
database design, as well as exploring new software and reporting bugs. DD Division also
provides operation and system support for the central database hardware. Engineering applications are developed by the groups concerned. The LEP experience has
shown the advantage of using a dedicated machine and of storing data centrally where it
can be shared. Coordination of application development helped to minimise the inevitable
redundancy of both application software and data caused by decentralised development. LEP
Database service management also proved to be a non--negligible task. Although tools for
database design at present are primitive, the advantage of structured analysis and of an
efficient database design are becoming recognised. PC "database" products cannot be ignored. They are cheap, easy to use and in
many cases adequate for the users' needs. The migration of these databases to ORACLE, once
the volume of data becomes significant or the data is of public interest, must be
considered. The portability of ORACLE, which runs under VMS, VM and other mainframe systems as well
as on a large number of workstations and PCs has led to a rapid growth in the number of
remote systems accessing a central database. The need for management tools for distributed
database systems is becoming apparent. The breakthrough of the 1980's is the exploitation of DBMS based on the Relational
Model which greatly simplified the task of application development. It is predicted that
in the 1990's there will be extensions to the present Relational Model with enhancements
in the direction of 'logic databases' and 'object--oriented databases', two different
approaches proposed to include the semantics or meaning of the data in a relational
database rather than programming them into the application software or storing them in the
mind of the developer. Distributed databases, where data is physically stored in many different places, but is
managed as though it were contained in a single database, are still a subject of research,
but systems which allow the interconnection of databases across heterogeneous networks are
emerging. Distributed databases could ease the load on a central database, but must be
carefully planned to avoid other bottlenecks. Development in the database field is moving away from the DBMS software and moving
towards providing user--friendly interfaces and more powerful development tools.
Workstation packages for database application development are beginning to appear on the
market. These make use of the powerful graphics and windowing tools as well as accessing
local or remote DBMS and guide the end--user by displaying icons (pictures) and/or menus
which he selects with a mouse. Expert system software to solve complex queries accessing
relational databases is emerging. Database design requires a more rigorous method of analysis than the
entity--relationship model offered by the analysis tools on the market today. TEAMWORK,
adopted as the CERN standard, is a powerful tool for SASD (Structured Analysis, Structured
Design). Software for using NIAM (Nijssen Information Analysis Method) for database design
is beginning to appear, but as yet, no package spans both SASD and rigorous database
design. As database technology becomes more widespread, the availability of data becomes
increasingly important. It is no longer acceptable to close the database down during
several hours for backup or for data reorganisation. DBMS software designers are working
on ways of executing these tasks while the database is active. The comments of Bachy et al. ORACLE is evolving with the technological trends. A procedural language to describe the
integrity constraints between 'objects' is partially implemented. Procedures will be
stored in the database where they may be accessed by applications. SQL*NET (a component of
the planned fully distributed RDBMS, SQL*STAR) allows users to query distributed data and
to modify remote data even if no local database exists. ORACLE Corporation along with
other RDBMS vendors is working on a solution to the integrity problem of updating
distributed data. In order to satisfy the PC market, ORACLE has now been grafted to LOTUS
1--2--3, a popular spread--sheet facility. ORACLE is also moving towards the workstation
environment with an increasing number of ports. Development is continuing on their design
tool, CASE, to provide tools for structured analysis, database design, software
documentation and maintenance. A version for the VAX workstation is expected soon. Natural
language support for multi--lingual environments is in development. Within the coming
year, on--line backup of databases should be possible. During the next decade both applications and data will increase significantly,
consequently demanding more computer resources, both power and disk storage. Database
transactions make heavy use both of computing power and of I/O processing, an important
consideration when aiming at providing an efficient database service. End--users, unlike
application developers, are intolerant of system failures and bad response times. An
increasing number of users will be accessing ORACLE from applications running on
workstations, which must be able to communicate with a database machine. The complexities
of the hardware connecting a terminal or workstation to the database computer must be
transparent. Application software development will also move to the workstation
environment. A public database service is provided on the general purpose VM and VMS systems. The
LEP database is installed on dedicated VAXs which are used exclusively for database
exploitation. The dedicated hardware solution adopted for the LEP database project gave
LEP Division the freedom to provide the service that the project required. Initially the
LEP machines ran independently. Today these systems are integrated into the Computer
Centre VAX cluster with the public service VAXs as well as the EUCLID CAD/CAM and other
service machines. The LEP DB machines are shared by all LEP users both for development and
production. The advantage of sharing data and applications is offset by the disadvantage
caused by competition for resources between the users themselves. The need for new
software and enhancements must be carefully weighed against the need for stability. As
members of a large VAX cluster, there is inevitable interference from other Computer
Centre services and from changes and upgrades which are not necessary to the database
service and which may introduce interruptions. There is a strong request from the machine divisions to provide a database service 24
hours a day, seven days a week. Applications concerning radiation protection, LEP
beam--monitoring equipment management and the LEP alarm system use the LEPDB for storing
data which is assumed to be permanently available. The LEP control system is less
sensitive to database availability, since although control data must be available for
loading into local storage before the start up of the machine, immediate access to the
database during a run is not essential. Down--times of under 30 minutes are believed to be
acceptable. Decoupling the database service from the public services and introducing a 'production'
VAX cluster may improve availability. Further measures could be taken such as scheduling
service interventions outside critical periods, extending the hardware maintenance
coverage and providing redundant hardware. Interference between users as well as
unpredicted hardware, software and network failures may, however, still occur. It may be
more efficient to build local safeguards against database unavailability rather than to
make significant investment in improving a service which, because of its generality,
cannot be guaranteed. Reliability and availability could be improved by isolating all or
part of the service. However, this may not satisfy the needs of the application. Two levels of support can be identified for database application development.
With the growth in complexity of database technology and the rapidly increasing number
of users, applications and data, the demands on Central Support for database
administration, consultancy, ORACLE expertise and exploration of new trends will continue
to grow. Consultancy includes recommending appropriate development tools, database design
software and optimisation methods. New trends such as storing "pictures",
graphics and documents must be explored as well as expert system technology and
workstation interfaces to ORACLE. The organisation or recommendation of training courses
for database design and application development using ORACLE is an essential part of the
service. Complementary support to that provided by DD Division is already provided by a small
team (3--4 people) in LEP Division who work in close collaboration with Central Database
Support. The support covers
Service Management Management of the database service includes the supervision of user
accounts, databases and exploitation of the service, prediction of growth rates for data
storage and CPU usage, estimation of the hardware additions necessary to meet the
expansion, and collaboration with DD Division to ensure optimum working conditions for
database users. The development of engineering applications is open--ended. Since most applications are
developed within the User Divisions, it is logical that the development support be
situated in the User Divisions where priorities for development, manpower and service
expenditure can be directly determined. The LEP Database hardware has evolved from one VAX/11--780 to two VAX 8700s in 1988 for
database application development and exploitation. An average of 720 KSFr. has been spent
per year on hardware, software and maintenance. The yearly expenditure can be roughly
broken down into 300 KSFr. for maintaining the present service and 420 KSFr. for
configuration evolution. Over the past six years, the capacity of the LEP database
machines has been doubled every two years. Although the emphasis has moved from the
development of planning and installation applications to control and maintenance, the
growth rate of approximately 3 to 4 account creation requests per week is continuing at a
steady rate. There were over 400 user accounts registered in mid--1988. Development
expenses have been kept to a minimum. However, with the rapidly increasing number of
workstations and PCs in the laboratory, a development budget must be included in the
future cost of the service. D. Jacobs All of the tools and methods described in this report require a high level of expertise
on the part of the users if maximum benefit is to be drawn from them. The required
knowledge falls into three broad categories:
With some reason it can be expected that staff are already equipped with the first kind
of knowledge by virtue of their education and experience. Design methodology, on the other
hand, requires special attention in the training program. Colleges and universities are
now training their students in effective use of CAE tools but at CERN, where new hiring is
rare and most of the engineering population graduated before CAE tools were introduced,
there is a great need to organise this kind of training. This also applies to Management
who supervise the work and must understand the new methods in order to see that they are
applied. Training in the use of particular tools is an obvious need, recurrent with each
new tool or tool version, although the learning time will tend to decrease as staff become
more familiar with CAE culture. Underlying all this is a need, in many cases, to provide foundation courses in computer
use. For many, the application of a CAE tool is their first contact with a computer and
basic training can help to avoid the kind of hurt surprise that greets the first crash of
a disk for whose files no backup has been made. Training must be adapted to the needs of the students and to the constraints of the
subject being taught and so there is no universal formula. The training service is well equipped to organise courses but, apart from those on
basic computer use and perhaps some on the operation of specific tools, it is probably
better that instructors should come from outside, thus ensuring a flow of new ideas into
the organisation. The more general courses may be organised at regular intervals but for the use of
specific tools it is necessary to respond rapidly to the needs of designers. Video--taped
courses are worth investigating for this purpose. There is a general feeling that it is most effective to follow concentrated courses
given on a number of consecutive days, during which the staff are relieved of their normal
work responsibilities. This permits them to invest the necessary time in revision and
preparation, essential to reinforce the course work. In the same vein, although running the courses at CERN may be the most economic
financial solution, there are benefits in taking the students out of their normal working
environment, which does not necessarily mean that they have to travel long distances. A
successful recent example has been the practice of running management courses at Ferney
Voltaire or Cartigny. Most CAE training will require hands--on use of the tools in question. This can pose a
particular problem for the training service in the case of workstation--based or
individually licenced products. All available copies of the tool will normally be heavily
committed to the laboratory's development programme and it will not be easy to release and
move them to a central location for the course. In such cases there is no alternative to
the expensive option of sending students to attend courses given by the supplier's
training organisation. The situation is obviously easier for software which is accessible
on the central computers or which runs on standard workstation or PC platforms with which
the training service can be equipped. Training is not free and training in the use of sophisticated tools can be quite
expensive. Supervisors and Management in general must come to realise, however, that to
deny their staff systematic training in the use of CAE tools is a false economy. It is
just not good enough to hope that they will pick things up as they go along. A typical
course might cost SFr. 2500. A week's wasted design time due to the use of inadequate
methods will cost the organisation more than the price of the course. The courses should be concentrated and it may, in some cases, be advantageous to
organise them off--site. Supervisors should be aware of their duty to make sure that their
staff follow such courses, not just for their own good but also for that of the
organisation. In the context of the studies relating to Computing in the Nineties it was agreed that
the MIS Board, with representatives from each Division, was the obvious mechanism for
coordinating the MIS contributions. It was decided to form three sub-groups for the purposes of specialist discussion and
provision of papers:
In addition each Divisional representative was requested to provide an individual
contribution for their Division. An extended MIS Board (with P. von Handel and E. Freytag as participants from DESY) was
then asked to consider the specialist input and in particular to reach consensus where
possible on the recommendations. The following papers were produced by the working groups and are referenced in the rest
of this text:
THE COMPUTING ENVIRONMENT FOR THE CERN PHYSICS PROGRAMME
Location of Physicists and Computing Facilities
Off--site and On--site Computing
The Role of National/Regional Centres
Private Facilities at CERN
Data Flow in HEP Experiments of the 1990s
Elements of Computing for CERN Experiments
Total CPU
Type of CPU
Mainframes used as scalar processors
Mainframes used as vector processors
Parallel computing
The Move towards Personal Workstations
Data Handling
Networking
EVOLUTION OF CENTRALLY ORGANIZED COMPUTING AND NETWORKING AT CERN AND AT REGIONAL
CENTRES
Introduction
Evolution of Batch Capacity
CERN Computer Centre Batch Facilities
Competitive and complementary services
Operating systems
Cost comparison of Cray and IBM--style computing
Manpower implications of the Cray service
The multi--purpose CERN VM service
IBM--compatible systems
DEC batch capacity in the CERN computer centre
Model for evolution of the central batch services
end 90
end 91
Cray
32
32
32
90
DEC
6
10
10
10
Total Capacity
74
102
99
154
Large Regional Centres
Other Batch Capacity
Emulators and parallel processors
Private batch capacity
Batch capacity in university physics departments
Workstations as batch processors
Evolution of General--purpose Interactive Facilities
Workstations at CERN
The Future of Central Time--sharing Services
Interactive Data Analysis
Networking Implications
On--Site Networking Infrastructure
Short term evolution
Medium term
Wide--Area Networking
Short term evolution
Medium term
Data Storage
Mass Storage Evolution in the CERN and Regional Computer Centres
Distributing the Data
On--site distribution
Distribution of DST data to external institutes
Storage Management
EVOLUTION OF CENTRALLY ORGANISED SOFTWARE SUPPORT AND DEVELOPMENT
The Changing Landscape
Operating Systems
Languages
Software Support for Experiments and Packages/Libraries
Support for Experiments
Packages and Libraries
SOFTWARE SUPPORT: DATA BASE SYSTEMS
Data Base Applications in Present Experiments
Addresses of Collaboration Members
Electronic Mail Addresses
Experiment Bookkeeping
On--line Data Bases
Description of the Detector Geometry
Calibration Constants
Event Data (or Memory Management Systems)
Bookkeeping of Program Versions
Histograms and other Physics Results
Software Documentation
List of Publications
Other DB Applications
Present DB Applications in Experiments: Conclusions
Broad Areas for DB Applications and their Requirements
DB Applications where Generalized Packages are Possible
Experimental Administrative Database
Electronics and material Bookkeeping
Tape and Production Data Base
DB Applications for which Tools could be Specified and Written
Detector Description
Calibration Constants
Data Base Systems: Recommendations
Education and Training
Design Support Team
Data Model Software
Portability of the Data Base Information
Area of Experimental Administrative Data Bases
Area of Bookkeeping Data Bases
Area of Documentation Data Bases
Area of Detector Description and Calibration Constants Data Bases
Area of Interactive Data Analysis Data Bases
Data Base Systems: References for detailed reading
SOFTWARE SUPPORT: SOFTWARE ENGINEERING
Analysis and Design, Methods and Tools
Historical Background
Situation in the HEP comuunity
Large Software Projects at CERN
Software Projects that have not Used Formal SE Methods
Organizational Tools
The Problems
Conclusions from Experiences with SE Methods
Requirements of Outside Institutes
Other Coding Support Tools
Source Code Management
Other Implementation Tools
Software Engineering: Recommendations
To Project Management
Training
Support and Consultancy
Software Tools
Licences
Requirements of the Outside Laboratories
SOFTWARE SUPPORT: GRAPHICS
Graphics in HEP
Graphics Applications
Implications for Software and Hardware
Graphics Software and Interface Standards
Basic Picture Generation
Basic Library Functions
Distributed Graphics
Document Processing
Graphics Hardware
Graphics: General Conclusions
SOFTWARE SUPPORT: USER TRAINING
Introduction
The Physicist's Demands
Documentation and Examples.
Formal Training Courses
New Working Practices for Physicists
Personal Workstations
PAW - The Physics Analysis Workstation.
Software Topics.
A Central Workstation Training Facility
Videa and Computer--based Training
TECHNOLOGY PREDICTIONS FOR THE 1990s
Introduction
CPU Technology
High--End Processors
Low--End Processors
Data STORAGE Technology
Disks
Solid State Devices
Tape/Cartridge
Optical Storage
Digital Audio Tape Technology
Exotic Technology
Summary of Data Storage
Data Transport
Local Area Network Capabilities
Wide Area Network Capabilities
Final Comments on Technology
APPENDIX
Current Mass Storage Usage in the CERN Computer Centre
VM/CMS
Disk Space
Users and Accounts
VAX/VMS
Disk Space
Users and Accounts
Cray
Summary
Accelerator Computing Requirements in the Nineties
Introduction
Main Conclusions
Background and History
General Model of Accelerator Computing
Workstations
Networking
Databases
Accelerator Design and Modelling
Computer Aided Design
Local Computing Centres, Servers, Central Support.
Control System Specific Issues
N-manufacturer policy
Support and User Feedback
Resources
Computing for Engineering
Composition of the Working Group.
Introduction.
Support for Microprocessor Users.
Introduction.
Hardware Support.
The current activity is a good model and should continue in the same spirit in the
future.
Software Support.
In the software area, which addresses both the development facility and the
run--time installation, two classes of environment need to be supported; a microprocessor
operating system with native development tools and a host--target configuration.
Training
Resources.
Summary of Other Recommendations.
Computer Aided Mechanical Engineering and Related Fields
Foreword.
Introduction.
Present Situation.
Near and Far Future.
Integration.
Population.
Hard-- and Software.
All the future upgrades are expected to be based upon a mixed (CI/LAVC) VAX cluster
configuration, the workstations being VAXstations (VS315 and successors) and lower price
stations supported by MDTV (MAC II, SUN,...) and the central computer, used as
number--cruncher and central database, a multi--processor VAX.
The 2--D drawing packages (AUTOCAD and EUCLID--IS--2D) should be usable on the
multi--purpose mid--range workstation which will be found on everybody's desk (the last
release of AUTOCAD is available on @mVAX, MAC II, Apollo,...). The two packages should be
fully interfaced to each other.
Structures and Resources.
Structural Analysis.
History.
Present situation.
Development into the 1990s.
User population.
Programs for structural analysis.
It is not the intention of the working group to recommend here explicitly and on
the basis of the actually installed versions, one or other of the programs that are
presently available. They all have their merits, are known to their users and cannot be
abandoned within short time.
Bridges should exist between the domains of structural analysis and field
calculations. Standardised data formats used for description of the problem (e.g. meshing)
and for the output of results would permit the transport of data from one aspect of the
problem to the other. The availability of source code for "peripheral" programs
could help the mutual adaptation.
Efforts must be made to profit from the interactive modelling facilities offered by
CAD systems (e.g. EUCLID) and of the possibility to implement a CERN--wide design
database.
Computing.
User support.
The central support group in DD division should be given the necessary resources
and should then be made responsible for installation and updating of a set of
"standard" and general--purpose S.A. programs, for documentation and for basic
user information. This work must be co--ordinated with that of staff members in user
divisions who should be appointed to train new users and to give technical advice.
Estimates of resources.
It may be supposed that the needs of the regular users for workstations or powerful
personal computers will be satisfied by their divisions. Computing on mainframes would be
done at the computer centre. There the work should be facilitated by permitting temporary
access to large amounts of disk space during calculations (up to 300 Mbytes or even more).
As was requested in the ACCSE report
Software Support for Field Computations.
Introduction.
Existing Field Computation Programs in use at CERN.
Static fields in 2 dimensions (2D).
Static fields in 3 dimensions (3D).
Electromagnetic cavity fields and others.
Present and Future.
TOSCA.
ANSYS.
Budget.
It is estimated that an annual central budget of about 100 kSFr. will be required
for software maintenance and for the purchase of new packages.
Human resources.
One staff member in the DD central support group should be assigned to look after
package installation and maintenance, advise on workstation acquisition and disseminate
information on new developments. He should work closely with the staff member assigned to
support of structural analysis packages so that some level of mutual backup is achieved.
The DD central support group should of course negotiate licence agreements on advantageous
terms and organise training in an efficient and cost effective way. Modern sophisticated
tools are complex to learn and their diversity should thus be limited.
Hardware.
The software makes use of the normal computing infrastructure at CERN, including
the computer centre and networking. In addition it will be necessary for divisions to
invest in colour--graphics terminals, PC's and workstations according to needs. Some feel
it to be desirable that newcomers to the field should make their first experience with PC
versions of the programs, since large amounts of central computing time can be wasted in
the hands of novices. It is clear that demands on the computer centre will increase, both
in terms of CPU time and disk space.
Analog Electronics CAE/CAD.
Introduction.
Fields of Application.
Integrated Circuit Design.
Discrete Component Analog Circuit design.
System Design.
RF and Microwave design.
User Population.
Conclusions and Recommendations.
Private versus collective access to CAE/CAD.
Designers feel that for designs of small to medium size, or for preliminary studies
of a concept, they need a private CAE/CAD working environment based on a PC or graphic
terminal, with locally multiplexed moderate quality printing and plotting. For more
comprehensive design work, a high quality workstation can serve 2 to 5 intensive users.
Apart from this, highly specific local working environments, composed of integrated
measuring or handling equipment and associated driving software is increasingly needed,
particularly for IC and RF design. In all cases, where compatibility allows, access to
common central facilities should be provided by networking.
Organisation of activities.
In the context of the present organisation of CERN, support activities are proposed
at three levels:
Some hints for foreseeable investments.
Education and human resources.
Last, but by far not least, there is a consensus that training on the subject, both
at the technical and the managerial level, has a capital role to play in introducing and
applying CAE/CAD successfully. Joint activities with outside lecturers and the CERN
training services should be organised.
Digital Electronics CAE/CAD.
Present Situation.
Future Trends.
Recommended Approach.
Resources for Digital CAE/CAD at CERN for the 1990s.
At present, no single tool can cover all requirements for CERN. Therefore, the
proposal is to provide and support 4 categories of tools. [Ref.]
describes the tools and [Ref.] the related resource requirements.
Management Issues.
Investments in modern tools will only be efficient if:
Training.
After an initial training (e.g. 1 week), designers will go through a learning curve
of 3--6 months, before attending advanced training (e.g. 1 week). After that, designers
should attend advanced training courses once per year (1 week). The recommendation is that
industry provides the training.
Infrastructure.
The essential elements of an appropriate support infrastructure are informatics and
operational support, co--ordinated library development and application assistance.
Design Management.
It is recommended that workstations be organised in logical clusters with
experienced CAE engineers designated as leaders of these "design clusters". The
leaders, besides working on a specific project, provide the applications support for
designers who (temporarily) join a cluster to perform their design tasks.
It is suggested to expand the use of CAE tools and methods in phases. The first
phase should see (i) the upgrade of existing Class--2 tools, (ii) the organisation of a
pilot design cluster, (iii) the co--ordination of the use of Class--3 tools and (iv) the
implementation of the basic infrastructure.
A final recommendation is to introduce a cost accounting scheme for design and
layout work. Such a scheme is a powerful tool for management to control in--house design
efforts through pricing, when services are transferred between different units. The
opinion of the working group is not unanimous on this last recommendation.
Benefits.
Database support for Engineering.
Present Situation.
Advances in database technology.
Development of ORACLE.
Since there is no significantly better product on the market and because of the
considerable ORACLE experience acquired at CERN, ORACLE will be the RDBMS offered to
engineers, at least in the foreseeable future. However, research into new DBMS products
and related software is essential.
Engineering Database Service Hardware.
The present LEP database service should be extended to the CERN engineering
community on dedicated VAX computers. Additional resources are necessary to provide the
computing capacity necessary for the larger user community. Possible ways of improving
availability should be investigated such as decoupling the database service from the busy
general purpose VMS service to reduce interference from non--database activities. The need
for a 24 hour service should be studied and, if the database service availability is
inadequate, alternative solutions proposed.
Database exploitation support.
Central Database Support.
The present Central Database Support team of 6 provides expertise and consultancy
not only to the engineering community but also to the MIS and physics database users. To
continue providing this service they should be reinforced by at least two posts to meet
the demands due to the rapid increase in database exploitation foreseen over the next five
years.
Local Application Development Support.
Until the future organisational structure of the Machine Divisions is known, the
support presently provided under LEP Division management should be extended to include the
whole engineering community. To cover the larger number of users, the present team of 3 to
4 should be increased to 5 in the near future and to 10 over the next five years, to cope
with the growing number of requests for application development. (At least 40 people
within LEP Division users groups are responsible for application development.)
Budget for the database service and exploitation.
Growth trends are difficult to predict. Extending the scope of the present LEP
database service to serve CERN's engineering community would clearly increase costs. A
minimum of 1.2 MSFr. per year is estimated to provide this service. approximately 300
kSFr. for maintenance, 100 kSFr. for small developments, 200 kSFr. for disk and cluster
hardware, and 600 kSFr. for configuration evolution. Major configuration changes tend to
be biannual. It is considered appropriate that the cost of workstations is covered by the
users.
Training.
Topics for training.
Organisation of training.
The duty to train.
Summary of recommendations.
The organisation should make available to its engineering staff training courses of
a number of kinds:
Bibliography.
MIS Computing in the 90's
Introduction
The following papers are indispensable input to the above:
This paper aims at extracting as briefly as possible the essential elements from the above papers. Detailed explanations and justifications are given in the original papers.
The general CERN MIS context is not very different from that of any other enterprise: corporate data are generated at the base, in a fashion which should be reflected in distributed electronic data entry at the location where the data are generated. These data should then be collected in large corporate data bases for correlation. Subsequently, managers and office workers should have access to these data from their office workstation to obtain decision information, to present them and to use them in models and reports at a local level. The current situation at CERN, its problems and the implications of recommended solutions are discussed below.
There are two main areas to be considered in CERN MIS activities: applications concerning corporate data and office systems. The data must be available to many at the same time, and must be a correct representation of reality. Office systems are used by MIS users in their daily work, combining corporate data and other information for management decisions. Following is an inventory of the current situation and its problems.
The CERNADP computer at the Administrative Data Processing center is a small IBM
mainframe
For reasons explained in
A limited number of office type applications have been used for a long while on the central mainframe computers. For many visiting scientists, the central machines are the only tool they have to help in office type tasks. Central text processing, mail and document storage is important.
The oldest application of computers in the office is for text processing, and while substantial progress has been made since the introduction of stand-alone word processors, the greatest needs remain located in this area: coherent laboratory-wide use, document interchange, the integration of graphics, document storage and retrieval, mathematical Text processing is a tool used daily in almost all MIS areas as well in physics.
Many vendors offer more than text processing functions on their equipment and in the
absence of standards, the situation is chaotic: the survey
Fortunately, at present, no new applications are undertaken for the obsolete systems, and over time they will be gradually phased out. Four contexts will however clearly remain important at CERN in the next years and all effort is concentrated in these "Four Operating Environments" (or FOEs): IBM with VM/CMS, DEC VAX with VMS, Olivetti IBM PC compatible with MS-DOS, Macintosh with MacOS. These are linked by "the fifth force": the Ethernet local area network.
It should be pointed out here that there is a growing number of high-end workstations with physics applications on which some MIS type work is done: for example, there are about 100 Apollos running a Unix-like operating system, which are partly used for document preparation. (note: during the next five years, we may see an increasing number of presentations on top of Unix as the underlying operating system kernel. This will be true for all powerful workstations but is also likely for the PC compatibles and Macintosh. The OS/2 for PC compatibles will probably make an impact in the MS-DOS world.)
As to personal computers, they came into CERN from mainly two directions: the experiments (Macintoshes) and accelerator control & engineering support (IBM PC compatibles). We observe that the middle ground, where users are not constrained by the needs of either of those environments, has been taken largely by Macintoshes, which are the most popular micros at CERN.
Finally, there is the corporate publishing unit. CERN has at all times had the need to distribute information about itself, and this unit has provided two functions: (a) the preparation of documents and the stipulation of their formats and (b) the actual production (bulletin, Annual Report, "Yellow" reports, ...). Around 20 document types are currently produced by Publications service, and 70 million A4 pages printed per year. At present there is no integration between the system used in the Publications unit and the others.
This section describes present user requirements and industry trends.
One of the most urgent problems is that of reducing the maintenance activities which now make up the overwhelming part of the day-to-day program of work, to allow progress towards unification and improvements. It is extremely desirable (mandatory?) that all major corporate data bases be unified in the next two to five years by replacing the current incoherent applications by a coherent and extended set (including for example the stores catalog, stores management and the inventory). Additionaly, divisional systems (buildings in ST and the User Office in EP) should be developed in coordination with MIS. The consequence of the heterogeneous state of affairs described in the previous chapter is a tremendous overhead in programming interfaces between systems that were not made for coexisting. The ADP application programmers spend most of their development time in programming new interfaces when the need for new applications arises.
A most important part of the requirements for the corporate data base engine is the set
of tools available for software analysis, development, implementation and maintenance. A
truly integrated data dictionary is essential
The applications should be easily available within the MIS user community, while the specialist functions should be available to Finance and Personnel divisions and well-protected for security reasons. Security and discretion is imperative, but where possible should not inconvenience users. Present security levels should not decrease.
A number of services external to CERN should be accessible: external commercial data bases, the data bases available through Minitel, a computer aided translation service, data bases useful to the CERN legal service.
A friendly system is needed at all levels. The user interface should tend to the
graphical, iconic interface
The various applications should be available from within the same context (no switching into subsystems with overall different philosophy) and should themselves provide a uniform environment. The user should be able to seek help without disrupting the context. These requirements are only fully met by hiding all environments except that of the user's workstation, which is now becoming achievable within emerging local area network environments. Thus the user should no longer have to be aware of the operating environment on which the corporate data bases reside, nor of how the data actually get to the screen of his workstation.
The system must be reliably available during office hours, but preferably also outside these hours. The loss of data is not acceptable. Its response should be fast. For "trivial" tasks, the response should be particularly quick. A trivial task is a task that the user considers trivial. The system should allow backups to be made without affecting the functionality. The hardware architecture of the data base engine should be extendible to accomodate growing loads. It should also have some (limited) non-stop operation features eg. disc mirroring.
A coherent office system should be available throughout the Laboratory. Two major facts will make this very difficult to achieve: (a) there is a community of physicists with needs different from those of the administrative MIS user, (b) the four operating environments will be here for at least another 5 years. For document preparation, the visiting physicists want compatibility with whatever hardware and software they are running in their institute while also wanting to communicate with resident CERN staff. The large installed base of the FOEs excludes the introduction of a new, panacea solution except at very high cost.
Among our users there is a great variety of educational level with respect to computing. This in itself creates great stress on the unit that has to provide service for both the sophisticated systems programmer (who knows little about administration) and the sophisticated administrator (who knows little about computing).
In the FOEs, the central VM/CMS system is unsuited for interactive work of the type now common on workstations, it is incompatible in its ways with all other systems. Basing the office system on mainframe computers makes insufficient use of the power of the workstations installed, and indeed makes them into very expensive terminals. However, workstations and even workgroup servers do not provide today adequate data storage technology for the volume, speed and reliability needed for a large shared document store which must be at the basis of the office system.
In any case, there is not enough manpower available in the present phase of CERN history to provide a good office environment on all of the FOEs. Thus it is indispensable to find a correct rôle for each of the FOEs without wasting resources on duplicating facilities, especially as, depending on the environment, this cannot be done effectively and efficiently.
Document production will increase dramatically in sophistication: users today want
excellent text processing tools, but only a few systems currently available are up to the
task, whereby each usually is strong in a few areas and weak in all the others. The search
for an acceptable set of tools for scientific document preparation is one of the major
tasks of OCS
It is highly desirable that coherent links between the office systems and the CERN Publications unit are brought into service and that the unit uses techniques compatible with those of the office systems.
The increases in interactivity of computing implied by the above list demand high bandwidths available everywhere, from the central data repository (which will have to handle many transaction-type requests simultaneously), through networks to the desktop workstation which will have to be a powerful graphics machine. It is in this bandwidth boosting activity that we see a potential major rôle of departmental systems.
All the MIS services should be accessible from the user's single workstation on his desk. This requires a high-quality, fast multi-tasking machine. Such machines are now available at affordable cost (eg. Macintosh II or Olivetti M380), the objective should be to select a minimal number of types and provide a maximum number of different services on the workstation. The availability of the workstation has brought us the problem of how to use it efficiently:
The following patterns are important to MIS work in the coming years:
MIS wants to
Thus the following services are necessary:
MIS should integrate the access to corporate data, transaction processing and electronic document handling with the use of personal workstations, which have text processing, spreadsheet and graphics applications locally. As much as possible, the local processing power should be used, processing should move towards more central systems only where necessary, for reasons of reliability and availability. A general guideline should be the use of standards wherever possible and acceptable. We propose therefore the following recommendations, which are discussed in more detail below (each explanation also has details of the resource estimates given in the table at the end). They are listed here in three categories, but all are equally vital.
A.1 Audit, modernise and subsequently automate the organisation's procedures
The document
A.2. Establish the CERN corporate data model
The purpose is to ensure a coherent approach to the creation of new data bases wherever needed. This is considered to be a prerequisite for the audit of A.1., and a refreshed model will be the result of the audit.
A.3. Use a unique DBMS if possible
This could be Oracle, which is supported by a vendor-independent company on all of the FOEs, but it could turn out to be different, as Oracle is not the only DBMS with this characteristic. Oracle is however already firmly implanted at CERN in other domains.
A.4. Replace all corporate applications with a coherent and extended set
This will have two advantages: it will diminish the effort now needed to run a heterogeneous mixture of systems and it will make it possible to move towards a truly homogeneous interface for the users. It is the equivalent of the reduction of the number of systems in office computing.
A.5. Integrate the existing platforms (FOEs)
Servers will be introduced to act as link between the Macintoshes and the PC compatibles. The central system is to be used as a document store, a backup store and a gateway node for some of the electronic mail exchange. It is not to be used as an interactive office system. The networks are Ethernet and Appletalk, as is already the case.
A.6. Introduce a comprehensive electronic forms handling system
A rudiment now exists and is used extensively in PS division. We should look for a commercial product or replace the existing system by a CERN-wide one. Its exact form again depends on the outcome of the audit. It may not be necessary to use electronic mail (or equivalent) if distributed applications can gain automatic access to a central data base engine.
B.1. Evaluate the use of a front-end system for ADP
Data would be put there for public use, which will give easy access to a wide range of users via a common DBMS, while leaving the back-end repository of corporate data as the secure system accessed only by the specialist departments. A study should be carried out in '89, at the end of which a decision should be taken on extending the approach.
B.2. Study EDI and introduce for contacts with suppliers
Commercial products are becoming available to help with electronic interchange of business documents. The body for standardisation of the forms has however not yet published more than a few of them, so that we have a breathing space before we have to handle EDI. Evolutions have to be followed closely.
B.3. Study the potential rôle of workgroup servers & departmental computers
Among the more obvious uses of the servers are: printing, backup, mail, file store cache. The workgroup servers can in principle be any make, but for reasons of familiarity, uniformity and compatibility machines from the VAX range are strongly preferred. DEC clearly are ahead in integration of services in multi-vendor environments.
B.4. Recommend an environment for desk-top office use
The desktop workstations are Apple Macintosh II and Olivetti M380 machines, with, where needed, A3-size or colour screens, and local or group laser printers.
However, Macintoshes will certainly remain the most popular machines for the next 2 years: the Mac OS achieves a level of integration of the applications sold by different software houses that is not matched by any other system today. They are preferred by almost anyone not constrained by technical issues. Both Macintoshes and IBM compatible machines will be purchased by divisions with specific technical needs. Both will be supported, leaving us free to benefit from a substantial technological leap on either side.
C.1. Upgrade the ADP computer
An upgrade of the present computer centre is undeferrable. the machine needs to be replaced by a 5 to 10 Mips unit in order to be able to continue to provide the present services.
C.2. Move the ADP computer to the computer centre
The move to the computer centre has already been accepted, it will enable the ADP group to concentrate on its key functions of analysis and applications development. It should be done quickly, for economy reasons since it will reduce the overheads now incurred in operation, backup, power supplies and air conditioning, as well as making it easier to link to other machines. Extensions will be easier.
C.3. Run a computer shop activity
There already is a shop providing the service of channeling the purchases of desktop machines of both PC compatible type and Apple Macintoshes. The shop also provides counseling, runs a few public services, and deals with the technical contacts with our suppliers. It is an important part of the OCS activities and must be extended.
C.4. Increase and organise user support functions
User training must be organised. A day-to-day help consisting of "hand- holding", question answering and help is necessary. As an example, if every PC/Mac user has one question per month, and if s/he knows there is a help service, then with the present level of 2000 stations, that help service would get 109 questions per day. This rate is probably exceeded today: we not do see a catastrophe only because users help each other and because the reliance on workstations is still very low. Increased user support is also needed for the corporate ADP applications which will be more widely used than today. ADP users need to be supported with analysis of their needs for applications.
C.5. Contract out the repair and maintenance service for workstations
The repair and maintenance activities should be contracted out. With an installed base of close to 6000 units in 1993, each with a useful life of 5 years, showing 1 important failure per year during this lifetime, that implies 6000 failures per year or approximately 27 per working day. There will also be upgrades and modifications, probably of the same order of magnitude.
C.6. Phase out all non-recommended systems over the next 3 years
The existing secretarial systems will be phased out and replaced by workstations. Some of the data on those systems must be transported to the new stations, taking a non-negligible amount of time. There is however no guarantee that the functions of existing applications will be preserved in all cases.
C.7. Replace the stores management system
The now obsolete computers used in the CERN stores should be replaced by a new centrally sited and operated system.
The tables below summarise our estimates for each of the points cited in the recommendations. Estimates are very approximate.
A short explanation for each item can be found in 4.3, after these tables.
A. Introduce coherence and modernisation |
B. Evaluation of future choices / developments |
C. Operational activities | |||||
900* | 900* | 900* | 900* | 900* | 900* |
Special projects |
Base-level services |
Global Total |
? = conditional on outcome of study
* = financed by the user division(s), not by MIS.
Estimates are very approximate.
A. Introduce coherence and modernisation |
B. Evaluation of future choices / developments |
Special projects |
Base level services |
Global total (MIS only): |
A.1 Audit, modernise and subsequently automate the organisation's procedures
The cost of external consultancies is estimated at 250 KCHF for two years. During the analysis phase of two years, two MIS analysts must assist, and during implementation of three years one analyst.
A.2. Establish the CERN corporate data model
First analysis and implementation needs 2 analysts for two years, maintenance goes on for ever with one analyst.
A.3. Use a unique DBMS if possible
If Oracle is not chosen, a license for software will cost around 100KCHF. A site license for Oracle exists.
A.4. Replace all corporate applications with a coherent and extended set
Costs for replacements of the four major corporate applications is estimated at 100KCHF per application, adapting them to the CERN environment depends on the willingness of CERN to change procedures, but is estimated at 5 manyears per application migration.
A.5. Integrate the existing platforms (FOEs)
Two systems programmers for two years are needed to provide a set of systems solutions.
A.6. Introduce a comprehensive electronic forms handling system
Two systems programmers are needed for three years to implement or adapt a solution that is sufficiently comprehensive and elegant. Maintenance is at 0.5 manyears per year.
B.1. Evaluate the use of a front-end system for ADP
In the case of choosing a VAX, a 10Mips/10Gb system is evaluated at 1MCHF in 1990. Evaluation is one year by an analyst. Implementation and maintenance, independent of the choice, is estimated at 2 analysts forever.
B.2. Study EDI and introduce for contacts with suppliers
This probably needs half an analyst in the first three years, because of integration problems.
B.3. Study the potential rôle of workgroup servers & departmental computers
If the study results in a recommendation of departmental computers, they could be introduced at a sufficient rate as corridor machines, about 10 per year, and needing CERN-wide two support staff (technicians). The machines would have to be purchased by the user groups at approx. 50KCHF per unit (based on today's approx. price for a μVAX).
B.4. Recommend an environment for desk-top office use
They need CERN-wide two support staff (technicians). The machines have to be purchased by the user groups at approx. 7.5KCHF per unit. For Apollo's and Unix, we estimate of the order of two software engineering posts to support these users with their MIS needs, but currently there is no manpower available.
C.1. Upgrade the ADP computer
This urgent upgrade to a more powerful IBM machine is estimated at 2MCHF, including 15Gb discs and cartridge-based backup and archiving unit (replacement of old equipment).
C.2. Move the ADP computer to the computer centre
The move will cost 300KCHF.
C.3. Run a computer shop activity
The activity costs three staff (technicians and data aid) not counting user support (see elsewhere), 50KCHF is estimated for risks and overheads.
C.4. Increase and organise user support functions
Five manyears per year are required, from both ADP or OCS staff, and the consultancy office will need to be equipped to handle the questions (50KCHF initially, 5KCHF later on).
C.5. Contract out the repair and maintenance service for workstations
This supplies work for 6 to 12 technicians, or approx. 900 KCHF per year in wages. However, it represents an extra cost of only 200 CHF per unit over its entire lifetime (not counting the price of the replaced or installed parts). One contact person is needed half-time. The cost of repairs is for the users.
C.6. Phase out all non-recommended systems over the next 3 years
Manpower is 2 programmers for three years, there is probably some small cost in special equipment and software during the changeover, in the order of 50KCHF.
C.7. Replace the stores management system
The system costs (estimated at 1MCHF) will be carried by ST division, MIS needs two analysts for two years to interface the new system to the other corporate systems.
This report has been prepared by the CERN Technical Board on Communications (Telecom) [Footnote: J.V.Allaby (Chairman), W.Blum, B.Carpenter (Editor), F.Fluckiger, D.Lord, C.R.Parker, G.Shering (Secretary).] in response to a request from the Steering Committee for the report "Computing at CERN in the 1990's" chaired by J.J.Thresher. Due to the very tight timescale, this report was first drafted on the basis of recent and current studies, and on the basis of formal and informal contacts with the CERN user community. Input from the LEP experiments was solicited during the summer concerning high--speed links. The interim version has been freely circulated as a draft since 16 June both inside and outside CERN, and all comments received have been taken into consideration. However, this report is not and should not be read as a statement of requirements for HEP [Footnote: High Energy Physics] in general, nor for on--line control and data acquisition systems. It refers only to the needs and priorities for general purpose networking at CERN. Neither is this document intended to prescribe a technical implementation plan; it is limited to stating the requirements. The general theme is one of continuity and expansion. Since the studies by Working Group 2 of the LEP Computing Planning Group [1--2], and the studies by ECFA SG5, around 1982, CERN has been pursuing a consistent data communications strategy [3--7], which we do not propose to change, but rather to reinforce, to meet future needs. Furthermore, input was solicited from members of ECFA Sub--Group 5 on Links and Networks, and the draft report was extensively discussed at the SG5 Meeting in September where three Telecom Board members were present. These sources and other basic reference documents are listed in an Appendix. Finally, an open presentation and very lively discussion took place in the overcrowded CERN DD auditorium in October. As a result of these and other interactions we feel mandated to stress the following points. Networking between HEP laboratories and CERN can offer the large collaborations unique possibilities in running and evaluating experiments, to monitor experiments from home institutes and to promote distributed computing. Bandwidths of 2 Mbits/s at the time of the LEP data taking are considered extremely useful. A variety of initiatives are being undertaken in several countries in order not to miss these opportunities. One of the most basic problems of experimental institutes, how to concentrate enough effort at CERN inside the limitations of travel budgets and teaching obligations, may be eased considerably using high quality network connections. These institutes rely on CERN to fulfill its role not only to provide functioning accelerators but also to provide services in the field of computing and, in particular, of networking. Computer networks need planning, management, development, and transfer of knowledge. CERN is the most important HEP institution in Europe from which the external HEP institutes expect to receive such services.
By 1990, CERN will have been at the forefront of data networking for twenty years. One can cite OMNET (1972), CERNET (1978), and CERN becoming a test customer of the Swiss X.25 service (1981) as older examples. More recently, CERN has become one of the world's largest Ethernet users (probably the first to have a site--wide bridged Ethernet), a major hub for the use of X.25 and DECnet in HEP, and the Swiss national node of the EARN and EUNET networks. CERN operates a unique set of gateways for electronic mail and file transfer. The reasons for this continual lead are those of fundamental necessity: the needs of a large, increasingly computerised, site, the increasing size and internationalism of the experimental collaborations, and the increasing size of experimental data samples. Our experience is that however much capacity is provided, it will rapidly be filled, and that there is always a suppressed demand for extra capacity. The typical annual growth rate for the traffic in each data communications service operated at CERN is now between 20% and 100%. The current pressure to decentralise data analysis (see the section on Central & Distributed Processing in the report on Computing for Experiments) will only emphasise this trend. The current illustration of this is of course the LEP experiments, whose preparation has depended critically on the availability of networking (principally for electronic mail in terms of human transactions, and principally for file transfer in terms of bulk network capacity). It is unlikely that the collaborations could have advanced as they have without these facilities. The MUSCLE report [8] has shown, furthermore, that today's network capacity is certainly inadequate for the data taking and analysis phases of the four experiments. [Footnote: An equivalent analysis of the needs for the new generation of American experiments leads to a similar result. ECFA SG5 believes that these two analyses can legitimately be extrapolated to HERA.] We cannot see this demand for more and more network capacity going away in the next few years. Indeed the availability of high bandwidth WAN [Footnote: Wide Area Network] capacity from European PTTs, the imminence of very high bandwidth LANs [Footnote: Local Area Networks] , and the approaching deregulation of European telecommunications, will encourage and intensify this demand. In addition, the range of people using data networking on the CERN site will increase precisely as the use of computers for non--traditional purposes increases. Thus the demand for more capacity per user will be multiplied by the increase in the user base. At the same time, the organisation of networking for European research in general will improve, moving from the present somewhat fragmented situation to a more integrated one (catalysed by the eventual arrival of standards). CERN will benefit from this process, but will continue to have some requirements which go beyond those offered by the emerging standards. A final background remark is that current networks can be described as "third generation" (industrialised and largely standardised). A fourth generation of networks, providing truly distributed computing rather than bilateral communication between computers, is now coming to maturity. In addition to continuing to provide adequate capacity and basic services, CERN must not miss the boat of distributed computing.
This chapter gives our view of the computing activity at CERN and in its external partners' institutes, as a scenario within which to state requirements for general--purpose networking. We assume that there will continue to be one very powerful Computer Centre at CERN. There, one will find most of the batch capacity and non--volatile storage, an operations centre, and the focal point for networks. We assume that roughly half of the interactive computing activity involves this Centre (either by direct connection, or indirectly through intelligent workstations). We additionally assume that there will be several powerful independent computing centres elsewhere on the CERN sites, which will provide some batch capacity but principally be used interactively for physics analysis. These will rely on the Centre for reliable non--volatile storage. Examples of these centres are the Work--station Clusters being set up by the LEP experiments. Each such centre would have a single well--defined connection to the general--purpose network. We also assume the existence of several networks for on--line systems (control and data acquisition systems), which should each appear as a single, self--contained, entity for purposes of security and management. These networks would each be connected to the general--purpose network by an appropriate gateway system. (Although the requirements for such on--line networks are outside the scope of this report, we do not mean to imply that they necessarily require different technology, but only that they should be logically separated and could be separately managed.) We assume that there will be several thousand intelligent desk--top devices around the sites (personal computers and workstations). At the beginning of the 1990's there will be more than 3000 dumb terminals, but we expect that this number will decline as the number of intelligent desk--top devices increases. Many PCs (or terminals) will be in people's homes. External communications will be required with several comparable laboratories in Europe and the USA, and with hundreds of smaller institutes and university departments, mainly in Europe and the USA, but potentially anywhere in the world. Each of these external sites will be subject even more than CERN to its own pseudo--political pressures from governments, PTTs, and manufacturers, so may not be free to choose communications techniques at the behest of CERN. Connections required to some countries may be forbidden for purely political reasons connected with restrictions on technology transfer. We regard this as being prima facie incompatible with the scientific openness of CERN. A quantitative prediction of network traffic from these assumptions is difficult, especially since all past experience is that traffic expands to fill all available bandwidth, with compound growth rates of at least 100% per year. Also, all previous studies of data traffic at CERN have shown an atypical pattern in which the highest throughput is often required over the greatest distances. For the early 1990s we believe that the aggregate cross--site bandwidth should be about 100 Mbps (including safety factors), as foreseen in earlier studies. Technological trends in LAN technology suggest that this could become inadequate from about 1993, but it is too early to predict how great an increase will be required. For off--site networking, the potential demand is effectively infinite and traffic predictions as such are probably meaningless. Certainly in the period under study, the bandwidth will be limited by cost or technology, not by needs. [Footnote: The technological limit today is at around 2 Mbps for mainframe--mainframe connections and perhaps double that for Ethernet--Ethernet connections. Both of these limits will evolve rapidly in the next two years. At least one proposal for use of 140 Mbps capacity is being prepared at the time of writing. ] In this connection, the recent very detailed bandwidth requirements developed by the American HEPnet Review Committee [10] clearly show that even mundane use of networks requires high bandwidths. We believe that these estimates can readily be extrapolated to European HEP. The 2 Mbps links between CERN and the major external centres recommended by the MUSCLE report are certainly inadequate to allow true distributed computing over a wide area. Our estimates of the real requirements by 1993 are collected in [Ref.], together with the current figures for this year and the extrapolated numbers for the end of 1989. It should be noted that the large increases in peak backbone traffic correspond in 1989 to the start--up of 4 LEP experiments with links from the 4 pits to the computer centre as well as the start of LEP data analysis, and in 1993 to the transition to fully distributed computing. It is useful to point out that even if the bandwidth and connectivity requirements are met by the communications infrastructure, performance bottlenecks may arise at the interface to central computers. (There is anecdotal evidence of 9600 bps connections providing one character per second of throughput!)
The above forms the context for our view of CERN data networking. Scenarios for networking must not pre--suppose any particular choice of operating system etc., as networking solutions must function in a diverse environment over many years. However, it must be noted that some companies have proved better than others at providing "peer--to--peer" networking which requires a minimum of central management. This has clearly driven many purchasing decisions in the HEP community in recent years, except in the pure Computer Centre environment where the cost of batch and storage capacity takes priority in the choice of suppliers. Any networking scenario must provide mechanisms for user relations, both on a day--to--day and on a strategic basis. We believe that daily user support tasks are best integrated with the general user support mechanisms of the CERN Computer Centre. Advisory bodies which are representative of the user community are also required. These bodies already exist, in the form of the Telecom Board for CERN's internal affairs, and ECFA SG5 for the European HEP community as a whole.
This chapter gives a brief summary of the services ideally to be provided by internal and external networks. It should be noted that user groups can and should build specific application--oriented services on top of these basic services. The services are listed in no particular order of priority, although the six more "traditional" services are listed first.
We note that technical evolution in areas such as fibre optics and ISDN [Footnote: The Integrated Services Digital Network to be offered in the 1990's by the PTTs.] continues without pause. We therefore expect that current 10 MHz LAN technology will be replaced by 100 MHz technology in the coming five years. Similarly, the current 64 kbit/second WAN technology will be replaced by 2 Mbit/second technology (indeed, this is already today largely a financial issue), and later by substantially greater speeds. As mentioned above, PTT regulations in Europe are rapidly becoming more liberal, and this process should be completed by 1992 with the establishment of the full Common Market. HEP must, if necessary, insist at the political level on its need for freedom from arbitrary PTT rules. We also note the rapid evolution in the area of standard protocols, and in particular OSI. [Footnote: Open Systems Interconnection standards defined by ISO and CCITT.] All technical requirements in this report must therefore be kept under continuous review and updated as conditions change.
Firstly we support the analysis of requirements made in 1983 in the LEP Computing Planning Group report. The resulting recommendations have been followed as far as possible by DD and still remain valid. We give here a summary of these recommendations:
Implementation of a general--purpose high--speed backbone network interconnecting LANs. |
Use of industrial products for the backbone, and integration with other facilities including digital telephone. |
Use of CCITT transmission standards across the site [i.e. G.703 series TDM (Time Division Multiplex) equipment and optical fibres]. |
These recommendations have still not been fully implemented due to the lack of suitable (100 MHz) industrial products for the backbone, but an intermediate backbone network has been created using slower products. G.703 equipment has been widely installed, mainly by SPS/ACC for the LEP site, and it is used both for the digital telephone and for the intermediate backbone.
Adoption of an agreed set of high--level protocols. |
(See below for more specific requirements on this point.)
Expand wide--area network services. |
This expansion has been a continuous process since 1983, notably concerning EARN, DECnet, and X.25, and shows no sign of slowing down.
Choose and install a standard LAN. |
ISO 8802.3 (Ethernet) has been chosen and installation is well advanced, and should of course continue.
Adopt an integrated approach to terminal connections. |
This recommendation is valid, but has proved difficult to implement, and does not allow for the impact of personal computers. However, the Telecom Board has recommended use of Ethernet for terminal and PC connections, with the digital telephone as a backup for remote buildings or special cases. The impact of ISDN in this area remains to be seen.
A CERN policy on communications protocols was adopted in 1985 [3]. Particular requirements in this area are:
Strategically, the protocols used should be independent of any particular computer manufacturer and should be able to communicate with the widest possible range of external sites with the minimum diversity of protocols. |
This requirement would be best matched by a migration to OSI protocols. This strategy matches that of the academic and research networks in the member states and even the USA. The subset of OSI protocols chosen by CERN should be as far as possible the subset chosen by the member state academic communities as a whole.
While awaiting OSI products, or in cases where the performance penalty caused by a standard protocol is too great, CERN requires to use interim (de facto standard, or manufacturer's) protocols. However, diversity of such protocols should be resisted and no development effort should be devoted to such protocols. Home--designed protocols should be avoided. |
Sadly, numerous interim protocols are in use today due to slow industrial progress in realising OSI protocols. We wish to emphasise the very substantial manpower cost of this diversity. All moves to introduce new protocols should be evaluated with great care.
Protocol gateways continue to be required as a major tool in providing connectivity despite diversity of protocols, and as an aid for migration to OSI protocols. |
We support the continuation of the existing gateway programmes (GIFT [Footnote: General Internet File Transfer, a file--transfer protocol convertor implemented by a CERN--INFN--RAL-- Oxford--NIKHEF--SARA collaboration.] , MINT [Footnote: Mail INTerchange, a set of mail gateways implemented at CERN] , and several industrial gateway products). Additional gateways should be implemented if the need becomes apparent.
This section gives a number of more specific requirements.
Access to CERN at speeds of at least 2 Mbps is required from major HEP sites in Europe, to allow decentralised analysis of the LEP data, as soon as this analysis starts in earnest. Higher speeds (equivalent to those attainable on a LAN) will be required as soon as practicable, to allow effective use of workstations over geographical distances. |
This requirement was initially raised by the MUSCLE report, and is justified in detail in the report on Computing for Experiments. The links will in all probability be based on PTT fibre optic cables and on the same G.703 TDM standards as used on the LEP site. The potential demand for bandwidth on such links is potentially infinite, with 2 Mbps being the minimum level at which the connection ceases to be frustratingly slow at peak hours. The limiting factors will be technical availability (for some countries), tariffs, and bottlenecks at the computer interface (for some computers). Until now, CERN has always viewed its WAN connections as a logical star with CERN at its centre, and CERN effort has officially been confined to support of connections to CERN. We have informally stepped outside this role in several cases, such as by acting as the hub for DECnet, X.25, and electronic mail in the HEP community, and by acting as host for the Swiss national nodes of EARN [Footnote: European Academic and Research Network] and EUNET [Footnote: European Unix Network] . CERN and ECFA SG5 have jointly taken the role of representing HEP in both the RARE [Footnote: Réseaux Associés pour la Recherche Européenne] Association and in discussion with the Eureka project COSINE [Footnote: Cooperation for OSI Networking in Europe] . These activities have been informally known as 'HEPnet'. At the initiative of ECFA SG5, a proposal [Footnote: Authors: R.Blokzijl (SG5 chairman), B.Carpenter (CERN), H.Hoffmann (DESY).] has been prepared for the the HEP--CCC to put HEPnet on a more formal basis. In summary, the proposal is that HEPnet should firstly represent HEP interests in the international networking community (RARE, COSINE, EARN, EUNET, etc.), and secondly coordinate the management of whatever HEP--specific network facilities are required to complement the general--purpose networks. It is proposed to create a HEPnet Requirements Committee as the successor of ECFA SG5, and a new HEPnet Executive Committee to coordinate the technical work. It should be noted that the relationships between RARE, COSINE, EARN, EUNET, and various national and regional networks are very complex and in constant evolution. Nevertheless, to avoid duplication of effort and wastage of money, it is desirable for HEP to use these networks as much as possible. Only by taking an active role in HEPnet can CERN strike the correct balance between use of the general--purpose networks and use of CERN--managed networks.
CERN should take an active role in HEPnet activities, without losing sight of the priorities of the CERN experimental programme. CERN should also, in conjunction with HEPnet, collaborate with general--purpose network activities (RARE, COSINE, EARN, EUNET) as long as this is beneficial to the HEP community. |
The dramatic growth in services, traffic, and number of users in recent years has caused a great increase in the load of operational management of internal and external networks. This would become even more acute if the creation of HEPnet led to an increased commitment by CERN to support transit traffic. Today, no operational staff are assigned specifically to data communications and many routine jobs are carried out by systems programmers. It is planned to assign one or two operational staff (on day--shift only) to alleviate this situation. Corresponding technical tools must also be provided, and the use of appropriate service contracts should be investigated.
Organisation, staff, and tools for network operational management should be boosted. |
For historical reasons, some users still depend on CERN--designed and --implemented protocols. Pressure on support staff, and the availability today of adequate protocols from industry, requires correction of this anomaly.
Dependency on CERN--designed protocols should be eliminated as rapidly as possible, in favour of industrially supported protocols. |
Thus CERNET should be phased out in the coming years, to avoid future maintenance problems. (The same argument clearly applies to the old TITN networks for the PS and SPS control systems.)
The general--purpose LAN infrastructure should take into account specific needs of the large experiments, to avoid their needing to implement "private" extended LANs across the sites. |
For example, this specifically concerns fast links to the LEP pits, and high--speed connections between workstation clusters and mainframe computers. The detailed justifications are given in the MUSCLE report. We would in general discourage any solutions requiring special (or worse, home--made) hardware or software.
Network services must be fully integrated between LANs and WANs. |
CERN users commute regularly between CERN and their home institute; as far as possible they should see a 'seamless' network in which the difference between LANs and WANs is hidden. In principle, there should be a single LAN/WAN gateway at CERN for each type of service, to simplify security and accounting procedures, and to optimise resources. The old telephone exchange will be replaced by expansion of the digital exchange within two years. This will then clear the way for use of the next generation of PTT services, namely ISDN, when this becomes a financially and technically attractive alternative to X.25 and/or LANs (probably not before 1992).
The use of ISDN services on and off the CERN sites should be planned. |
It is clear that the need for data communications facilities will continue to grow dramatically, as in recent years, certainly until about 1992. Whether we then reach a period of relative stability, or whether growth continues, depends on the future of the CERN experimental programme. (For example, a trend away from high--statistics experiments might significantly alter the traffic patterns implied by the MUSCLE report). However, even in a period of stable demand, there would be a need for a rolling replacement programme for equipment purchased in the 1980's. We prefer to assume that the resources required throughout the 1990's can be extrapolated from the well--understood needs of the next four years. Our financial model is therefore built out of five items:
The expenditure on Ethernet and the backbone should tail off, but will presumably need to be replaced by other expenditures (such as the replacement of INDEX as it becomes unreliable, or the introduction of the next generation of LANs and of ISDN equipment). To allow for this assumption, a flat extrapolation of current rates of capital spending has therefore been adopted in [Ref.]. However, two capital items are not extrapolated beyond 1992, namely the STK digital exchange and the MUSCLE items. It is not excluded that by 1991, other major capital items will prove to be necessary, so a contingency line has been added. This financial model is coarsely rounded; it would be misleading to imply greater precision than is possible by giving more precise figures.
1. including 0.4 MSF paid by DA 2. including 3.7 MSF paid by DA Two major items are not included in table [Ref.]. The first of these is the cost of the high--speed external links required by the MUSCLE report. It is not clear whether these lines would be funded by CERN or by the external institutes. At normal PTT tariffs, a total annual operations budget rising from 500 KSF in 1989 to 12 MSF from 1992 would be required. Costs of conventional direct links to CERN are currently paid by the Institutes requesting them. The second item is the cost to CERN of establishing and operating its share of HEPnet, currently unknown. It is very difficult to give a true picture of staff requirements, since almost everybody involved in computing support is involved for some fraction of his or her time in communications support. The estimates here are only for people involved for a significant fraction of their time in communications. The Telecom Board gave a detailed staff scenario in its October 1987 report on resources [7], showing a total of about 20 FTE staff [Footnote: Staff figures are quoted as full--time equivalents, not as individuals.] supporting internal networking, and about 15 for telephony. About 28 of these FTEs are in DD today, and several more should be transferred to DD from SPS in due course. Several other Divisions have a few staff involved part--time in data communications, as do all large collaborations. To this one must add staff needed to support external networking (currently ~10 FTE). This gives a total of about 30 staff providing central data communications support in DD today. These staff are overloaded by a burden of installation work, service support, and user support, almost to the exclusion of any work in preparation for the future.
>Current Staffing Levels (FTE) for Communications | ||||||
EP/EF | All Divs | |||||
Software | 14 | 0.5 | 0 | 0 | 1 | 15.5 |
Hardware | 2 | 0.5 | 2 | 0 | 0 | 4.5 |
Technicians | 15 | 0.5 | 4 | 0 | 0.5 | 20 |
Operators | 15 | 0 | 0 | 0 | 0 | 15 |
1. all DD groups, not only CS 2. includes one ST telephone technician It is necessary for the other Divisions to make available to DD substantial technician support for the planning, installation, and commissioning of their Ethernets and for ongoing operational support. The direct participation of the large collaborations in the planning and operation of their on--line Ethernets has been assumed and is fortunately happening. It will be necessary for this participation to be extended to cover also the first--line intervention in the extensive local area networks planned for the analysis centres, where the changing topology demanded by the physics requirements will require local expertise, supplemented by the minimum of consultation with the central DD support group. For this purpose, the DD support group will have to organize suitable training courses in network management and operations, to allow physicists, engineers, and technicians to acquire the appropriate expertise. Furthermore, extensions of our wide area networking possibilities can only be made possible by increased collaboration and participation by the major European HEP institutes in their maintenance and operation.
>Future Staffing Needs (FTE) for Communications | ||||||
EP/EF | All Divs | |||||
Software | 17 | 0.5 | 0.5 | 0.5 | 1 | 19.5 |
Hardware | 4 | 0.5 | 0 | 0 | 0 | 4.5 |
Technicians | 18 | 0.5 | 0.5 | 0.5 | 1 | 20.5 |
Operators | 13 | 0 | 0 | 0 | 0 | 13 |
1. all DD groups, not only CS 2. includes one ST telephone technician Continued priority should be given to Communications, within the general staffing policy of the Organisation. In particular, we believe that the Communications Systems group needs several additional staff to help with daily administrative or operational tasks, today wastefully carried out by software engineers, and an additional boost in software effort. [Ref.] shows current approximate FTE staff, and [Ref.][Ref.] the numbers we consider are required to cope with future needs. They exclude on--line networks, TV, audio, and safety communications. [Ref.] only represents a survival level of staffing only; an increase of about ten posts could readily be justified, for example to provide the level of support that many users would like to see. [Ref.] also implies that all departures must be compensated, with the exception of some telephonists, and that operational and software manpower should be boosted by a few posts, as an absolute minimum for survival. It will be noticed that the future staffing needs have been slightly increased from those estimated in [7], since we now believe, as a result of our extensive consultations, that we were too modest in the former estimate.
Plus individual input or comments from: R.Dolin (L3), P.Frenkiel (Paris), F.Harris (Oxford), M.G.N.Hine, P.Ponting (CERN), E.Pagiola (CERN), A.Rouge (IN2P3), W. von Rüden (ALEPH), B.Segal (CERN), and several members of the DD/CS Group. Apologies to all those who have been overlooked.
The scientific part of the division consists, including visitors, of approximately 120 individuals at any given time; they are located in 60 offices. The many visitors have a very diverse background and stay in close contact with home laboratories. As a consequence they are involved with all major computer systems in Europe and USA; in particular short term visitors cannot be expected to invest time and effort into learning unfamiliar new systems which will not be useful at home, however excellent such systems may be.
In the past the research activities of the Theory Division only depended in a minor way on computing. The exceptions mainly concerned a few individuals associated with phenomenological work, and, otherwise, small programs. This pattern is now changing in different ways.
As a consequence of this generalized use which affects all members of the division independently of detailed research profiles we attach a high priority to the equipment of all offices with terminals, PC's and workstations (e.g., Mac II's) as well as ready access to printing facilities. At present, the TH equipment in this respect is far below normal standard in member states and the US.
The implications of these general considerations will now be quantified.
The use of electronic mail presently concerns less than 50% of the division and visitors. We expect this to grow to 100% in the next few years. The electronic mail in TH is predominantly international. Local electronic mail is mainly of collective nature (seminar notices, etc.); in addition a strongly extended information on preprints, etc., from the library is developing. Individual mail messages from the TH go predominantly to other laboratories. At present only a few modems at home exist. We believe the use of modems will increase substantially in the future, in particular since electronic correspondence is conveniently done after normal working hours from home.
Systematic use of databases and networking only concerns the division to a minor degree. The main exception is the library public service, with the distribution of information from the SLAC, ISIS etc. database on preprints, books and so on, both inside CERN and to other laboratories.
Our plan is to equip all of the 60 scientific offices with terminals in addition to the public terminal rooms. A special problem arises from the rapid turnover of visitors; only the simplest operational terminal system is easily used by short term visitors. As a consequence, personal computers will only be installed for TH staff and long term visitors. Furthermore, we must provide visitors with access to personal computers of varying types including printing facilities, so that they can use their home programs readily. Low end workstations will mainly be installed in public areas. We estimate needs for personal computers to be mainly Mac's, with a few Olivetti's and some Mac2's. The demand for workstations may be underestimated.
Estimated cost: a mix of 20 units at an average price of SFr 6.000: SFr 120.000 (including programs; in particular TEX). Printers: 2 laser printers: SFr 12.000 + operating cost SFr1.000/year.
For the choice of other terminals we would ideally like to have cheap graphics screens, in particular in view of the WYSIWYG demand for scientific text processing with TEX. For the same reasons there will be pressure for a few units with big screens for page editing. Present compromises are Falcon's and Atari's. Estimated cost: 35 offices SFr 2.500 = SFr 87.500.
For its internal needs, the library will install about 10 Mac's: cost 10xSFr 3.500 = SFr 35.000.
The TH divisional secretariat has presently an AES system as a networked multipurpose workstation. This system is poorly compatible with all the other divisional activities and should be replaced by a modern, suitable system in the coming years. This requires in particular special screens, but can probably otherwise be based on a medium size workstation suitable for text processing.
Estimated replacement cost: SFr 50.000 + 1 printer SFr 14.000.
In the TH division we do not expect to put workstations on everyone's desk even if the money is available. This is not suitable for short-term visitors who ask for the simplest mode of operation first.
Low end workstations in the TH division will be excellent for public areas, for the divisional secretariat and for some of the staff members and long-term visitors. We expect "dumb terminals" to be a non-negligible part of the standard equipment even in the future, complemented with good personal computers and cheap graphic screens.
On the software side, TEX is essential to TH both on VAX and on IBM; we will certainly use it also on Mac's and PC's when available. This is presently an unofficial international standard, that cannot be ignored for practical work in particular in collaborations. Any other standards must coexist with TEX for quite a while.
For drawing, plotting, etc., the general system like PAW is an overkill and too sophisticated for most TH work. We need quick, simple, standard systems (Mongo, Topdrawer). People who want to use such systems once in a while desire to invest only the briefest time possible before using them. For the average visitor, plotting is less trivial than DD thinks.
Of the presently available symbolic manipulation programs REDUCE is presently partially OK, but too slow (SCHOONSCHIP on ATARI's is faster, specialized and in most cases more suitable for theorists). The program MAXIMA now on SYMBOLICS is inadequate.
UNIX is presently of minor interest to the division although we have one happy user.
The problem with visitors is that nothing is ever achieved: they change rapidly. There is probably a need for some introductory training of visitors on arrival; one may consider a 3-5 hour practical introduction to mail and text processing. The courses should be given when needed in groups of, say, 6 to 8 after signing up on arrival.
We need the availability of good, simple, practical guides with down-to-earth instructions. This concerns in particular basic use of terminals, printers, etc., access to various sytems, mail and text processing. Practical examples are very important. We need in addition better availability of documentation on TEX and plotting. These should be readily available via the divisional secretariat.
The library has a special status both inside the division as well as inside CERN and the memberstate laboratories as a public service. Its operations presently are profoundly changed by the installation of an electronic integrated library system foreseen for 1989 at an approximate cost of 200-250 kSFr for computer hardware and another 75 kSFr for peripherals (input/retrieval stations, printer, barcode readers, etc.). The accompanying software will cost 100-150 kSFr. The optical disc project for preprints will be implemented in late '89 or early '90; the cost for the central equipment is foreseen of 250 kSFr in budget discussions. The exact impact of these electronic systems on the use by internal and external users is presently difficult to evaluate with any precision. However, any remote access station implies 15 kSFr per high resolution screen with PC. A minimum of 5-6 such screens must be installed in public areas. Individual groups at various places inside CERN may ask for private screens. To this should be added eventual network costs, so that image transfer of the optically stored articles can be made with little delay inside CERN. A substantial amount of printing will accompany this service (say 500.000 to 600.000 pages/year).
In the past, access to external databases via the CERN networks has sometimes been difficult. The goal should be that such consultations can be made directly by the R & D personnel without the intervention of intermediaries in SIS.
Equipping TH with terminals: 20 Mac's, PC's, Mac II's: 120 kSFr 35 graphic screens, etc.: 87.5 kSFr 2 printers: 24 kSFr Secretariat: Replacement of AES System: Workstations, screens, printer: 64 kSFr Library: Computerized library system: Hardware: 200-250 kSFr Software: 100-150 kSFr Peripherals: 75 kSFr Optical disc storage system: 250 kSFr High resolution large screens (5 to 10) 75-150 kSFr
Normal replacement and maintenance at 15%/year should be added as well as operating costs.