hep experiment integration within griphyn/ppdg/ivdgl rick cavanaugh university of florida...
TRANSCRIPT
HEP Experiment Integrationwithin
GriPhyN/PPDG/iVDGL
Rick Cavanaugh
University of Florida
DataTAG/WP4 Meeting
23 May, 2002
23.05.2002 VO Management 2
Lawrence BerkeleyNational Laboratory
BrookhavenNationalLaboratory
Indiana University
Boston University
ArgonneNationalLaboratory
U Michigan
University ofTexas atArlington
OklahomaUniversity
Grid credentials (based on globus CA)
distributed
• Process of updating to ESnet CA
credentials
Grid software: Globus 1.1.4/2.0, Condor
6.3 (moving towards full VDT 1.x)
ATLAS core software distribution at 2 sites
(for developers) (RH 6.2)
ATLAS related grid software: Pacman,
Magda, Gridview, Grappa
Testbed has been functional for ~ 1 year
Accounts (individual user, group) created at all sites
GRAT – Grid Application Toolkit for ATLAS grid applications
(RH 7.2)
US-ATLAS Test Grid
23.05.2002 VO Management 3
Develop a Condor+GDMP+Magda package• Magda: distributed data manager prototype
• Data production is waiting for cataloguing hooks using Magda Develop data analysis tools (to simplify user experience)
• Enhance GRAPPA web portal Use Virtual Data Toolkit (VDT) and test the GriPhyN Virtual Data
Catalog (VDC) Participate in Data Challenge 1 Automate grid package production mechanism Deploy a hierarchical GIIS server Develop an MDS information provider for Pacman-deployed software Interoperate with US-CMS Test Grid and EDG
• Run ATLAS apps on US-CMS Test Grid (done!)
• Run ATLAS apps from US-ATLAS Site on EDG Testbed (done!)
Near Term US-ATLAS Plan
23.05.2002 VO Management 4
UCSD
Florida
Wisconsin
CaltechFermila
b
Princeton
Master Site
Remote Site 1
IMPALA/BOSS
mop_submitterDAGManCondor-G
GDMP
BatchQueue
GDMP
Remote Site NBatchQueue
GDMP
Grid credentials (based on globus CA)
distributed
• Process of updating to ESnet CA
credentials
Grid software: VDT 1.0 Globus 2.0 beta Condor-G 6.3.1 Condor 6.3.1 ClassAds 0.9 GDMP 3.0
Objectivity 6.1 MOP – distributed CMS Monte carlO
Production
Testbed has been functional for ~ 1/2 year
Decentralised account management
DAR – Distribution After Release for CMS
applications (RH 6.2)
US-CMS Test Grid
“MOP”
23.05.2002 VO Management 5
Near Term US-CMS Plans
Prototype Virtual Data Grid System (VDGS)• Based upon VDT (and the GriPhyN Virtual Data Catalog)• First prototype by August• Production prototype for November
Grid-enabled Monte Carlo Production• Build upon the CMS and MOP experience (already quite mature)• Run live CMS production this Summer• Integrate with VDGS for November
Grid-enabled Analysis Environment• Based upon web services (XML, RPC, SOAP, etc)• Integrate with VDT and VDGS for November
Interoperate with US-ATLAS Test Grid and EDG• Run CMS apps on US-ATLAS Test Grid• Run CMS apps from US-CMS Site on EDG Testbed
23.05.2002 VO Management 6
D0 SAM Deployment Map
Cluster data according to access patterns Cache data which is frequently accessed Organize requests to minimize tape
mounts Estimate resources for file requests
before they are submitted Make decisions concerning data delivery
priority
All sites are functional D0 centers that routinely send/receive data to/from FNAL
• anticipate one or more stations at each collaborating institution eventually
Processing Center
Analysis site
23.05.2002 VO Management 7
Commissioning of SAM for CDF
GOALS
•support 5 groups that do data analysis•enable access to datasets of interest•production availability of the systems•limit impact on CDF enstore
CDF portakamp 6509
CDF Offline 6509
Border Router
CDFen
DCache
fcdfsam
sam stationenstore stat ~1TB
CachePerm.Disk
fndaut (sun)
name servicesam db server
optimizerlogger
web servermonitoring
fcdfora2
Oracle DB(dev, int)
nglas09
sam_station(analysis) Cache
Fermilab
remote
sam_station(analysis) Cache
remote
sam_station(analysis) Cache
remote
sam_station(analysis) Cache
remote
sam_station(analysis) Cache
remote
sam_station(analysis)
Cache
STKen
5TB
100 MB
1 GB
100 MB (multiple)
fcdfora1
Oracle DB(prd)
CD Switch
STATUS
•Hardware and Software infrastructure in place
•Translation of the CDF DFC ready to go in production
•Developed AC++ interfaces to SAM to retrieve and analyze files. Automatic output to SAM not ready, yet.
•Enabled access to DCache.
•Deploying to test sites to sort out configuration issues.
•Test user are starting now to use SAM to do physics.
23.05.2002 VO Management 8
Conclusion
Other non-HEP experiments (LIGO, SDSS) not mentioned in this talk
LHC Experiments have short term plans which are• aggressive
• Test Grids are still young and fault-prone• Inter-experiment and inter-grid integration• Distributed data analysis• Distributed Monte Carlo data production
• but realistic• Uses existing (for the most part) software and tools• Emphasis is on building prototypes and learning from them
FNAL Experiments appear well integrated ! Critical need to demonstrate the value of the grid!