Possible scenarios for LHC data analysis

1
Possible scenarios for LHC data analysis

• All analysers will live at CERN
• Valuable fraction of the data analysis will be
  done at PNPI and around (it means PNPI has to
  behave itself as regional leader)
• PNPI computing facility will be used only
• PNPI will use all the available computing power.

2
LHC Data analysis at PNPI and around

• In total we should have in 2005 the following
• 50K-90K SpecInt95 of processor power
• 2-5TB disk cache
• 10TB-50TB tape library (tape robot)
• All the equipment could be used in different
  manner
• In one place (for example at PNPI)
• In several places in SPb and Gatchina.

3
As soon as possible HEPD needs for computing

• We should have as soon as possible
• Small tape library (10 cartridges) with drive DLT
  8000 (about 10K)
• 0.5TB disk cache (about 10K)
• 1K SpecInt 95 (about 10 dual processor 1 GHz)
  (about 20K)
• Network Equipment and Accessories (about 2K).

4
Existing and planned channels at St.Petersburg
region
5
HEPD urgently needs for at least 2 Mbit to SPb

• Currently available link from PNPI to SPb is
  reasonably below critics
• That means PNPI needs to upgrade the link to SPb
  (not the Internet)
• I would like to emphasize that partner
  Universities at SPb do need for the same (good
  channel in between Universities and PNPI).

6
BCFpc hardware (picture)
7
Projects for distributed computing

• Main goal of all projects
• To realize more or less uniform access to
  distributed data, computing power, other
  resources
• GriPhyN Grid Physics Network (www.griphyn.org)
• DataGrid (grid.web.cern.ch/grid)
• RRCF (www.pnpi.spb.ru/RRCF)

8
Completed tasks in year 2000 (1)

• Seminar at SPb University (A. Chevel, March
  2000)
• CERNlib and related software was installed at
  IHPCDB (A. Chevel, T. Serebrova)
• Published paper on LHC computing at Russian
  journal Byte Russia number 2, 2000
• Meetings at PNPI on LHC computing (see
  www.pnpi.spb.ru/RRCF all people from CSD).
• Upgrade of the existing computing farm
• (A. Lodkin, A. Oreshkin)

9
Completed tasks in year 2000 (2)

• It was created testbed setup for system Globus
• Globus was installed on pcfarm and Linux
  station
• The people from HPCDB was asked to install
  Globus on spp.csa.ru (the installation is in
  progress now).
• It was installed client part of AFS on pcfarm.

10
Conclusion

• HEPD needs In nearest future (2001?) to upgrade
  the link to SPb.
• HEPD needs to realize in nearest future (2001?)
  the 10-20 processor computing farm.
• The End