Download presentation
Presentation is loading. Please wait.
Published byKimberly Ray Modified over 8 years ago
1
Western Tier 2 Site at SLAC Wei Yang US ATLAS Tier 2 Workshop Harvard University August 17-18, 2006
2
Western Tier 2 was approved in July 2006 SLAC has a Open Science Grid deployment for years Has worked on testing ATLAS environments since 2005, on existing resource Successfully run ATLAS jobs via Panda/Grid and local submission New equipment will come before December. Will provide service to ATLAS in September.
3
The Western Tier 2 Team at SLAC Management ( 0 FTE on ATLAS fund) Richard Mount 10% (time) Chuck Boeheim 15% Randy Melen 20% Advisory Board Technical Support ( 1 FTE on ATLAS fund) Wei Yang Western Tier 2 Contact 30%-100% ATLAS has the highest priority Booker Bense Grid infrastructure 20% Lance Nakata Storage 20% Scientific Computing and Computer Services 30%
4
Resource for Open Science Grid Primarily used by ATLAS 4 SUN V20z in production Dual Opteron 1.8Ghz, 2GB memory, 70GB local disk OSG 0.4.0 Gatekeeper / Gsiftp (for DQ2) GUMS 1.1 support US ATLAS VO VOMS 1.4.0 / VOMS Admin 0.7.6 500 GB NFS Space for OSG $APP, $DATA, $TMP on dedicated Solaris server 20 VA Linux 1220s for development
5
Resource for ATLAS MySQL replica for CondDB DQ2 site server / Web proxy for Panda Pilots 500 GB NFS space for DQ2 data 10 job slots per user to LSF batch system for grid users Access to LSF for local ATLAS users AFS space for ATLAS software Kit and environment 250 GB work space for local ATLAS users Prototype dCache
6
“Shared” Resource Leveraging existing Infrastructure and Expertise ~ 3700 CPU core LSF batch nodes, RHEL 3 and 4 ~ 30 CPU core interactive nodes, RHEL 3 and Scientific Linux 3 10 Gb/s to ESnet, 10 Gb/s to DOE data-intensive science, 1 GB/s to Internet 2 Expertise on OS, batch, storage, network, security, power, cooling, etc. RedHat provides very low cost license and excellent support
7
Challenges: Grid jobs overload LSF by checking job status frequently Cache job status information Batch nodes have no internet access Web proxy for Panda Pilots Local CondDB replica JobTransformation 11.0.X.X doesn’t use CondDB replica iptables NAT rules on batch nodes and redirect TCP traffic Security issues with DQ2 web server and MySQL Want to use LSF fair share instead of dedicated queues
8
Challenges, cont’d Prototype dCache Admin, PNFS, and pool nodes in Internet Free Zone Gsiftp and SRM doors have Internet access Pool nodes on Solaris 9. Get very useful info from user forum Using RHEL instead of SL provides significant benefit to SLAC SLAC ATLAS physicists will provide validation for running ATLAS code on RHEL What hardware to buy next? How much memory? Storage type? dCache NFS xrootd …
10
Future Plan 67% funding on storage 33% funding on CPU power Western Tier 2 is capable of providing more CPU to ATLAS due to resource sharing A Western Tier 2 web page to provide info about What resource is available, how to access Links to CERN, BNL for more generic info about ATLAS Provide user support for the Western Tier 2 via BNL RT system RT-RACF-WTier2@bnl.gov
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.