Grid Computing Status Report

1
Grid Computing Status Report

• Jeff Templon
• PDP Group, NIKHEF

NIKHEF Scientific Advisory Committee 20 May 2005
2
HEP Computing Model

• Tier-0 measurement center (CERN)
• Dedicated computers (L2/L3 trigger farms)
• Archival of raw data
• Tier-1 data centers
• Archival of 2nd copy of raw data
• Large-scale computing farms (e.g. reprocessing)
• Spread geographically
• Strong Support
• Tier-2 user facilities for data analysis /
  Monte Carlo

3
Worldwide HEP Computing Needs
4
Amsterdam Tier-1 for LHC

• Three experiments LHCb / ATLAS / ALICE
• Overall scale determined by estimating funding in
  NL
• Contribution to experiments scaled by NIKHEF
  presence321
• Resulting NIKHEF share of total Tier-1 needs
• LHCb 23
• ATLAS 11,5
• ALICE 5,75

5
Amsterdam Tier-1 Numbers

• Status GOOD!
• Basic collaboration with SARA in place
• Attitude adjustment needed (response time)
• Appropriate funding line in NCF long-term draft
  plan
• Just enough concerns about me-too (grids are
  popular)
• Community-building (VL-E project)
• Pull me-too people into same infrastructure

6
Overall Status LHC Computing

• LCG a successful service
• 14,000 CPUs and well-ordered operations, active
  community
• Monte Carlo productions working well (next slide)
• Data Management a problem
• Software never converged in EDG
• May not be converging in EGEE (same team)
• Risk losing HEP community on DM
• Makes community-forming (generic middleware)
  difficult Ill just build my own, this one
  stinks

7
Results of Data Challenge 04

• Monte Carlo tasks distributed to computers across
  world
• Up to 3000 simultaneous jobs per experiment
• 2.2 million CPU-hours (250 years) used in one
  month
• Total data volume gt 25 TB

For LHCb NIKHEF 6 of global total

• See it in action
• Backup

8
Transport of primary data to Tier-1s
?
9
The Dutch Contribution
LCG Service Challenge II
10
Local Status

• Positioned well in LCG EGEE
• Present on blessed Tier-1 list
• One of best run sites
• One of first sites (3 in EDG, compare 4 in WWW)
• Membership on
• Middleware Design Team (US collaboration here
  too)
• Project Technical Forum
• LCG Grid Applications Group (too bad, almost
  defunct)
• Middleware Security Group
• Etc etc etc
• D. Groep chairs world-recognized EUGridPMA
• K. Bos chairs LHC Grid Deployment Board

11
Local Status 2

• NIKHEF grid site
• Roughly 300 CPUs / 10 terabytes of storage
• Several distinct components
• LCG / VL-E production
• LCG pre-production
• EGEE testing
• VL-E certification
• Manpower 8 staff, interviews this week for three
  more (project funding)

12
PDP Group Activities

• Middleware (3 FTE)
• Mostly security -- best bang for buck local
  world expert
• Operations (3 FTE)
• How does one operate a terascale / kilocomputer
  site?
• Knowledge transfer to SARA (they have support
  mandate)
• Contribute regularly to operational middleware
• Applications (3 FTE)
• Strong ties to local HEP (ATLAS Rome
  production, LHCb Physics Performance Report, D0
  SAMGrid)
• Community forming LOFAR KNMI, looking for
  others

13
Industrial Interest
GANG Hosted _at_ NIKHEF IBM, LogicaCMG, Philips,
HPC, UvA, SARA, NIKHEF 16 industrial
participants (24 total)