MONARC plenary meeting, 15th March 1999
  • Participants

  • Introduction, H. Newman
    The introduction dealt on key issues related to Regional Centers and rised some relevant related questions which will be discussed further also with the representatives of possible Regional Centers.
  • Report from the LCB 10.3 Meeting, MONARC Shedule and Planning, L. Perini
    The issues dealt with in the slides include:
  • Simulation validation
  • Ideas on RC Guidelines
  • Review of Milestones
  • April Meeting of RC representatives
  • May Workshop (2 Days)
  • June Progress Report
  • Status of the Analysis WG, P. Capiluppi
  • Connection with Simulation WG
  • How proceed with choice of Models
  • Discussion
    One key point is that we have to parametrize the analysis model according to the simulation needs, and thus look at Iosif's tool. A set of different models is hard to specify before getting and interpreting simulation results for some first model.
    I.Gaines: we should try to consider also different strategies ( e.g. recomputing vs mooving); the differences between models are not due just to range of parameters.
    H.Newman: Architectures, Ranges and Strategies, Network and access profile have to be brought together; even without Objectivity details this is going to give significant results. Who is allowed to access data from tape? Who can do localization independent analysis?
    P.Capiluppi : have we to discuss a review of milestones, mainly internal ones (internal=not the ones put in the Schedule Chapter), in May workshop?
    H.Newman: even if some milestone may turn out too ambitious for the available developpment of the tools, what is important is that we get relevant results, clarify some critical points and make advancements in the understanding of the problems we face.
  • Status of the Simulation WG
  • K. Sliwa summarized the progresses achieved, on the basis of his presentation at the Analysis meeting of two weeks before.
    Between the points yet to be developped he mentioned a description of client-server interaction, and of the data model as envisaged by Analysis WG. The idea would be to be able to take into account in the simulation how often it will required to go back to the higher data structure.
    H.Newman noticed nobody knows on OO data, a data model and a prototype are required.
  • Simulation Demo
    Iosif Legrand showed his simulation Demo and announced the setting up of a web site with the various documentations, demos, etc. of the MONARC simulations he has built. From this site it is also possible to download the code. The demo shown is a run of the single center model with no randomness in, as can be obtained from the Web. People are encouraged to try the tool and give feedback, also on the most significant quantities to be plotted for interpreting the results.
    A frozen working version will be provided together with the one undergoing continuus development.
    In the discussion the follows, general apreciation is expressed for the amazing amount of work done by Iosif and for the progress achieved.
    I.Gaines underlined the issues to be adressed for the use of the simulation: we will have to identify which parameter have to be varied for specifying an architecture and analysis scenario, and which results have to be used for the choice; but now we have to start to use the tool.
  • Status of the Testbeds WG L. Luminari
    H.Newman asked what bandwith is assumed in the Testbed WG planning, and L.Luminari aswered it is the one available now.
  • Report of the Architecture WG, V. O'Dell
    The presentation dealt with the Report on Current Experiments Compared to LHC, on which MONARC was expected to release a document already some time ago.
    Some discussion followed:
    S.O'Neale : reconstruction onsite/offsite numbers for LEP are uncorrect, the same for analysis.
    K.Sliwa : the same for CDF.
    People are encouraged to read the document and react...
    In the report is mentioned as a "Failure" the fact that most of the computing was done onsite, due to poor access to raw data and s/w infrasctructure not up the date. However the two motivations can be traced back to the decision that the experiments made of doing most of the work onsite. This decision was well justified when it was taken; actually this cannot be considered a failaure except from our limited point of view of searching for proven model of distributed computing. The only computing that was largely distributed was MC: we plan to study further this point and to understand the motivations.

  • Summary of Steering Group outcome

  • We could discuss with the referee the possibility of redefinining some internal milestone in a more clear and less ambitious way, comptible with the overall goal and schedule, in the june report.
  • Need to worry on a very short time scale how to set up a real simulation team, able to discuss results and steps to get at them, and stop relying on Iosif alone. S.O'Neale to be asked if he can follow, possibly Morita too, even if he is more interested in testbeds. A small group of people from Analysis WG could meet with simulation team every two weeks for suggestions about results and tool facilities. A Web page should be created for collecting such suggestions without creating a flood of mail.
  • Regional Centers meeting confirmed with Lyon,FNAL, Berkeley, INFN, RAL, possibly someone from Germany, Japan, and all interested countries
  • For the Architecture WG documents, we have to take up again the tape point and the ratio between LEP forecasts and starting point and LEP final reality
    • 1st April 1999