(%204%20years%20experience%20)

1
( 4 years experience )
Joint CHAIN/GISELA/EPIKH Workshop, 10 December,
2010, Valparaiso, Chile
The scope of the event is to show both to the
scientific community and to the decision makers
the usefulness of e-Infrastructures and the
importance of their sustainability.
2
Outline

• History
• Cluster structure (local, Grid, )
• Workflow jobs, data, problems
• Conclusions and Outlook

3
History
4
Cluster structure
Rackmount Dell PE 2950, Xeon 1.6GHz, 2GB RAM,
140GB SAS Tower AMD 1.6 GHz, 512MB, 120GB IDE
Intel Core2Duo, 4GB, 2TB RAID SATA
Local
Grid
PROOF
master
PBS client
gLite-UI
UI
PBS server PBS nodes
lcg-CE gLite-WN
slave
CE
WN 1..11
xrootd
SE
POOL 1..3
Data
SE-dCache
MON
GPU 1..2
CUDA
Resource Center (RC) SL4, x86_64, gLite 3.1,
dCache 1.9.5 (gold)
5
Workflow jobs

• Local
• MPI, GPU, ATLAS offline, Minerva, tutorials

• Grid EELA
• prod.vo.eu-eela.eu (GAMOS etc.)
• oper.vo.eu-eela.eu (obsolete)

• Grid EGEE/WLCG
• ATLAS, NA61 (?)
• DTEAM, OPS

6
Workflow experimental data
7
LHC Data Grid Hierarchy
8
Network bandwidth
9
It works!
Grid CPU hours per country
10
Grid accounting
11
Conclusions and Outlook

• Lesson 1 team work!

• Lesson 2 network (bandwidth, firewall etc.)

• Upgrades hardware, software, network.

• Storage up to 100TB and more!

• New services (WNS, LB etc.) for NGI.

• Grid or Cloud?

Cooperative!
12
Cooperative work?
13
Cooperative work!
14
(No Transcript)