Minutes 5 October 06

• Phone:
  • TRIUMF: Denise
  • GridKA: Andreas
  • FNAL: Eric, Lee
  • ATLAS: Shasha, David Front
  • CNAF: Barbara
  • NDGF: Olli
• CERN:
  • 3D: Dirk, Zbigniew, Luis, Eva
  • LHCB: Marco
  • ATLAS: Gancho

• Update on phase 2 site status:
  • Expected phase 2 sites to report plans and concrete schedules. Dirk will send an email to contact them again.
  • Triumf (Denise by email): hw in place, power infrastructure ongoing. Setting up shared storage. Db nodes installed: RH4 64 bits. Wiki page is already updated.
  • NDGF: hw have been ordered. Dirk proposed a video conference to help on the database setup.
  • NIKHEF: hw have been ordered.
• LHCb and ATLAS COOL tests:
  • LHCb(Marco): Streams setup from online to CERN and from CERN to Tier1 sites ready. First test was successfull: 1 condition per second during the whole night (this is what is expected for LHCb)
  • ATLAS(Gancho):
    • delete operations using rowid (Florbela code) non scalable using Streams, rowid cannot be translated easily on destination. Primary key used instead shows good results.
    • materialized views: used on the source database. They are not replicated, transactions must be filtered.
    • delete one error from the exception queue takes long time.
    • waiting for Taiwan to join: they have problems with OCFS. Dirk proposed Barbara to be in contact with them because at CERN it is not used.
    • BNL network problems to be analyzed by the network people.
• CMS Frontier tests:
  • details on agenda
  • Luis:
    • queries from David Front for COOL compresion rate is too high - unrealistic
    • zip leves from 1 to 9: difference is not valuable. CPU consumption is similar (server)
    • test from FNAL: numbers are similar to Tier0 - dependant on the network
    • Squid + Gigabit at FNAL (prepared by Lee): available for tests