ATLAS TIER2: INFN Milano

1
ATLAS TIER2 INFN Milano

• Luca Vaccarossa
• I.N.F.N Milano
• luca.vaccarossa_at_mi.infn.it

2
Current site hardware status

• CE 25 WNs (dual xeon processors)
• 13 - 2.4 GHz 1 GB RAM public IPs
• 12 - 2.8 GHz 3 GB RAM private IPs (GW on CE)
• 2 CLASSIC SEs (front-ends for NAS servers)
• 1.6 (18) 1.6 (87) TB filesystems
• No SRM interface

3
Site Network status

• uplink to GARR network with bandwidth guaranteed
  34 Mbps and peak bandwidth 140 Mbps
• many elements are connected to a 10/100 switch
  (not SEs)
• 7500 cisco router (a new one Juniper is not yet
  in place)
• improvement of uplink in the (near?) future.

4
Manpower

• The plan of development of T2 (and engagement in
  SC3) is under elaboration
• Good Grid experience
• participation to the InfnGrid/LCG/EGEE grid
  production infrastructure
• Good will

5
Issues Data Management

• Problems with LCG data management tools
• lcg-gt calls the BDII and checks infos about a
  SE if the SE is down or the BDII is down, the
  command crashes. Used a workaround
• lcg-cp if the connections times out, the command
  hangs. Used a workaround
• lcr-cr On some failures, it can leave the file
  catalog in an inconsistent status. Moreover it
  hangs if the connection times out. Used a
  workaround for the time-out
• Due to these inefficiencies, we should have
  spread the data over more sites by hands
• Most failures due to DM problems
• Need for a reliable file transfer service.

6
QUESTIONS

• Where is going FTS client ? SE or WN ?
• Can FTS cohabit with standard file transfer tools
  ?
• Which kind of files will be used in throughput
  phase ?
• There is any danger for data already storaged at
  T2 ?
• How many (dedicated?) resources are needed in the
  throughput phase ? And in the service phase?