Available Resources

SPRACE Cluster

The SPRACE cluster is a T2 site of the CMS collaboration at the Worldwide LHC Computing Grid (WLCG) infrastructure.

General Summary

Production Time History

Upgrade Time History

SPRACE Current status

The SPRACE current status, after the FAPESP thematic phase V upgrade, that was installed by the end of May 2012, is

• 13.698 KHS06 of Processing Ressources
• 787TB of Disk Space
  • 30TB of Stage-Out Space
  • 250TB of Group Space (125 TB per group)
  • 200TB of Central Space
  • 170TB of Local Space
  • 127TB of User Space (~42 Users of 4 TB each).

Worker Nodes Summary

SPRACE has 144 worker nodes corersponding to 1088 computing cores. Those servers were bought at different times (phases), according to the evolution of the project, that started in 2004. Equipment of phase 1 were decommissioned in August 2010 because they are 32bit architecture and were no longer useful for CMS produtcion. Equipments acquired in phase V were installed in the end of May 2012, corresponding to more 64 workernodes/768 cores, enhancing significantly the computing power and storage capacity of the cluster.

(*) HEPSPEC06 benchmarks from https://www.gridpp.ac.uk/wiki/HEPSPEC06. Values for Centos 5.3 - 64bit, gcc 4.1.2, and same amount fo ram memory as in our servers. For each server with 2 x Intel Xeon Dual-Core 5130 @ 2.0 GHz the HEPSPEC06 is 30.95 and for each server with 2 x Intel Xeon Quad-Core E5620 @ 2.40 GHz the HEPSPEC06 is 118.30.

(**) Estimate for the HS06 value for the processor E5-2630 based in the SPEC CINT 2006 values for the processors E5620 (~29) and E5-2630 (~40), extracted from http://www.spec.org/cpu2006/results/cint2006.html .

Storage Summary

SPRACE has a dCache based storage with 787 TiB of effecive disk space, distributed in three SunFire (48TB raw disk space) servers, five Supermicro (72TB raw disk space) servers, and 4 SGI Summit + Infinite 2245 (135TB raw disk space) servers.

Head Nodes Summary

SPRACE has 5 head nodes, one for local users access (access), one for open sience grid compute element middleware (osg-ce), one for open sience grid storage element middleware (osg-se), and two for general tasks (spserv01 and spserv02).

Network Equipment Summary

SPRACE is connected to internet by a 10 + 10 Gbps link provided by ANSP. All head nodes and storage servers are connected by a 10Gbps to the internet. All worker nodes have a 1 Gbps conection to switches with 10 Gbps uplinks to the core switch.

Decommissioned Hardware

Some machines, bought at phase I of the project, were decommissioned because they were based on a 32 bits architecture. Other machines bought at phase II were also decommissioned because their hardware warranty has expired.

Processing hardware

Storage hardware

Phase	Vendor	Model	Raw Disk Space	# of units
I	Dell	PowerVault 220S	2TB	2
II	Dell	PowerVault 220S	4TB	2

Head Nodes hardware

Network Equipment

WLCG pledges

U.S. CMS Facilities

• U.S. CMS Tier-2 Facilities Deployment Status

CMS T2 Associations and Allocations

• PAG/POG Allocations at Tier-2 Sites

WLCG pledges for 2013

According to the WLCG pledges for 2013, a nominal T2 site is

• 10.6 KHS06 of Processing Ressources
• 787TB of Disk Space

WLCG pledges for 2012

According to the WLCG pledges for 2012, a nominal T2 site is

• 9.5 KHS06 of Processing Ressources
• 787TB of Disk Space

According to the Service Credit in 2012, a nominal T2 site is

• 10.9 KHS06 of Processing Ressources
• 810TB of Disk Space
  • 30TB of Stage-Out Space
  • 250TB of Group Space (125TB per group)
  • 200TB of Central Space
  • 170TB of Local Space
  • 160TB of User Space (~40 Users of 4TB each).