uk computing
Post on 30-Dec-2015
26 Views
Preview:
DESCRIPTION
TRANSCRIPT
12th November 2003 LHCb Software Week 3
RAL Tier-12002 - 312 cpus4 racks holding 156 dual 1.4GHz Pentium III cpus.
March 2003 - extra 160 cpus 80 x dual processor P4, 2.66GHz, Xenon
Dec 2003 - extra 400/500 cpu200 - 250 dual systems
Operating system now RedHat 7.3. Batch = PBS. CSF Legacy Equipment 250 cpus (450MHz - 1 GHz) Total ~1000 cpus
12th November 2003 LHCb Software Week 4
RAL Tier-1 Mass Storage
March 2003 - Extra 40TB11 new disk servers, each with 2 x 1.8TB filesystems
Dec 2003 - Extra ~70TB
2002 - 40TB26 disk servers, each with 2 x 0.8TB filesystems.
Disk Cluster
Total ~150TB
12th November 2003 LHCb Software Week 5
RAL DataStore
STK 9930 (Powderhorn) tape robot
June 2003Updated with 8xSTK 9940B drives.Transfer speed 30MB/sec each driveTape capacity 200GB
5,500 slots = 1PB potential capacity
Current capacity limited by number of tapes.
12th November 2003 LHCb Software Week 6
Switch_1 Switch_2
RS6000 RS6000RS6000 RS6000
fsc0 fsc1 fsc1fsc0
9940B 9940B 9940B 9940B 9940B 9940B 9940B 9940B
1 2 3 4 5 6 7 8
11 14 11 1415
fsc1fsc0fsc1fsc0
12 13 12 13 15
rmt1 rmt4rmt3rmt2rmt5-8 rmt5-8rmt5-8rmt5-8
A A A A A A A A
STK 9310 “Powder Horn”
Gbit network
1.2TB 1.2TB 1.2TB 1.2TB
12th November 2003 LHCb Software Week 7
GRID at RALLCGCurrently 5 worker nodes in testbed.15.8.2003 LCG On-Line10.9.2003 Upgraded to LCG1-1_0_022.9.2003 Upgraded to LCG1-1_0_1
Amount of future hardware deployed in LCG depends on experiments and GRIDPP.
EDGEDG 2.1 Deployed on development testbedEDG 2.0 On main production testbed.EDG 1.4 Gatekeeper into main production farm.
12th November 2003 LHCb Software Week 9
2003 Data Challenge 2004 (same sites)?Site SPECint2k*hour
s% Share
SPECint2k*hours
Storage(GB)
Bristol 5,520,705 0.9% 37,019,600 183
Cambridge
12,881,646 2.1% 86,379,067 428
Imperial 78,516,698 12.8% 526,500,978 2,606
Oxford 7,360,940 1.2% 49,359,467 244
RAL 54,593,642 8.9% 366,082,711 1,812
ScotGrid 47,232,701 7.7% 316,723,244 1,568
TOTAL 206,106,332 33.6% 1,382,065,067
6,841
Factor of 6.7
Still hope for ~10% share from 3 largest centres?
12th November 2003 LHCb Software Week 10
Tier-1 Resources for LHCb
Requested for DC04 (April - June):From Marco’s numbers (assuming same share as DC03)
CPU requirement - 366M SI2k*hours 6TB of disk for "permanent copy" of all DSTs (may reduce to 1TB if pre-selection is used) to be used for analysis. Existing disk servers (3.2TB) used to store MC production from RAL and other UK sites before transfer to tape/CERN.. Mass storage of 7TB to store SIM+DIGI+ data from all UK sites.
But actual resources will depend on competition from other experiments.
12th November 2003 LHCb Software Week 11
0
100
200
300
400
500
600
700
800
900
1000
UKQCD
Other
D0
Alice
LHCb
Atlas
CMS
BaBar
GPP-only
90%
Capacity
CPU Requirements (KSI2K)
LHCb
x3
LHCb
LHCb need~20% of farm for 3 months
12th November 2003 LHCb Software Week 12
UK Tier-2 Centres
NorthGridDaresbury, Lancaster, Liverpool,Manchester, Sheffield
SouthGridBirmingham, Bristol, Cambridge,Oxford, RAL PPD
ScotGridDurham, Edinburgh, Glasgow
LondonGridBrunel, Imperial, QMUL, RHUL, UCL
12th November 2003 LHCb Software Week 13
Tier-2
Number of CPUs
Total CPU
(KSI2000)
Total Disk (TB)
Total Tape (TB)
London 677 335 26 8
NorthGrid 815 263 43 28
SouthGrid 436 229 32 8
ScotGrid 259 126 18 20
Total 2187 953 118 64
Tier-2Number of CPUs
Total CPU
(KSI2000)
Total Disk (TB)
Total Tape (TB)
London 2454 1996 99 20
NorthGrid 2718 2801 209 332
SouthGrid 918 930 67 8
ScotGrid 368 318 79 0
Total 6458 6045 455 360
Existing Hardware(April 2003)
Estimated Hardware (Sept 2004)
12th November 2003 LHCb Software Week 14
Liverpool New MAP2 facility now installed
940 3GHz/1GB/128GB P4 Dell Nodes
SCALIManage installed, RH9 20 CPU CDF facility now
installed, Fermi RH Linux and 5.9TB disk
MAP memory upgrade (270 nodes)
EDG 2.0 being installed
10% of DC04 would take17 days on all processors.
Initial LHCb schedulingproposal:50% of farm for ~1 month.
~1.1M SPECint2k
12th November 2003 LHCb Software Week 15
ScotGrid
ScotGRID Processing nodes at Glasgow (128 cpu)ScotGRID Processing nodes at Glasgow (128 cpu)• 59 IBM X Series 330 dual 1 GHz Pentium III with 2GB memory • 2 IBM X Series 340 dual 1 GHz Pentium III with 2GB memory • 3 IBM X Series 340 dual 1 GHz Pentium III with 2GB memory • 1TB disk • LTO/Ultrium Tape Library • Cisco ethernet switches
ScotGRID Storage at Edinburgh (5TB)ScotGRID Storage at Edinburgh (5TB)• IBM X Series 370 PIII Xeon • 70 x 73.4 GB IBM FC Hot-Swap HDD
Phase 1complete
Phase 2 - now commissioningUpgrade database server in Edinburgh16-20TB disk storage in Edinburgh5TB disk storage in Glasgow (relocation from Edinburgh)Edge servers for EdinburghNew kit for Glasgow - CDF and eDIKT
12th November 2003 LHCb Software Week 16
ImperialViking at London e-Science Centre:Upgrade to ~500 cpu cluster(33% HEP + bioinformatics +…)Ready to join LCG1
Ulrik - “Factor of 5 seems realistic” running across~3 months.
Note: Other potential resources coming online inLondon Tier 2...Royal Holloway ~100cpuUCL ~100 cpuBrunel(BITLab)64 dual Xenon nodes + 128 more nodesTimescale? LHCb use?
top related