Date post: | 03-Jan-2016 |
Category: |
Documents |
Upload: | lawrence-gibbs |
View: | 218 times |
Download: | 0 times |
US ATLAS Grid Projects
Rob Gardner
Indiana University
Mid Year Review of US ATLAS Computing
NSF Headquarters, Arlington VAJune 20, 2002
http://www.usatlas.bnl.gov/computing/grid/
June 20, 2002Rob Gardner US Grid Projects 2
Projects Overview
Particle Physics Data Grid (T. Wenaus)
Year 2 of 3-year project to deliver vertically integrated grid services to experiments
Initial ATLAS activity: distributed data storage on the grid (Magda) grid analysis (new)
EDG kit evaluation and US ATLAS testbed certification (ANL); Monitoring, information servers (BNL)
GriPhyN (R. Gardner)
Year 2 of 5 year project: main deliverables for ATLAS: VDT, Pacman, Grappa; Chimera toolkit: virtual data
catalog and data language (VDC, VDL, VDLI)
Virtual data prototyping effort by S. Vanaichine (ANL) P. Nevski (BNL)
Architecture for virtual data portal developed (Argonne-Chicago-IU project)
iVDGL (R. Gardner), Networking (S. McKee)
Platform on which to design, implement, integrate VDT
Boston and Indiana University Prototype Tier2 Centers
Forum for grid interoperability – collaborate with EU DataGrid, DataTag, etc.
Networking: U of Michigan leads see Internet2 working group: http://www.internet2.edu/henp/
US ATLAS grid applications testbed (K.De)
Lots of contributions from various Lab and University groups! BNL,ANL,LBL,BU,OU,UM,UTA,IU
June 20, 2002Rob Gardner US Grid Projects 3
MAnager for Grid-based DAta
Designed for ‘managed production’ and ‘chaotic end-user’ usage
Designed for rapid development of components to support users
quickly, with components later replaced by Grid Toolkit elements
Deploy as an evolving production tool and as a testing ground for Grid Toolkit
components
Adopted by ATLAS for 2002 ATLAS Data Challenges
Developers - T. Wenaus and W. Deng (fulltime, postdoc)
PPDG Magdahttp://www.usatlas.bnl.gov/computing/grid/ppdg/
Info: http://www.usatlas.bnl.gov/magda/info
The system: http://www.usatlas.bnl.gov/magda/dyShowMain.pl
June 20, 2002Rob Gardner US Grid Projects 4
Status
Full time developer/support Wensheng Deng Development, support: ATLAS DC, US ATLAS testbed, LAr bench test DAQ, GDMP
integration
ATLAS DC production usage for cataloging (passive and active) and
replication Improved usability, doc for replication; dynamic replication; more replication modes DC1 data replication from BNL to CERN to begin To be used for data replication among the ~18 sites of DC1 phase 0
260k files, ~11TB cataloged at present
GDMP integration beginning Issues given to GDMP in Jan re: Magda integration fixed in V3 David Rebatto, a CS working for Laura Perini, is helping GDMP3 being set up in Milan; will install Magda and do integration work there
Torre Wenaus
June 20, 2002Rob Gardner US Grid Projects 5
PPDG Year 2 & Magda
GDMP integration
Use GDMP to support a ‘publish/subscribe’ replication mode within Magda, targeted at
‘conventional production’ usage (what GDMP is designed for)
Magda still ‘adds value’ with a more flexible approach to data management and replication
Application in demos
Build Magda outwards into distributed job management/analysis
Distributed job management is to be the ATLAS PPDG yr 2 focus
Together with distributed analysis?
Apply as replica manager in hybrid store common project
Build it outwards (or adapt it) to support more metadata? E.g. data history
information; HES metadata (HEMP a teeny step in this direction)
To be eclipsed by Alien? Examine Alien and evaluate
Torre Wenaus
June 20, 2002Rob Gardner US Grid Projects 6
Pacmanhttp://physics.bu.edu/~youssef/pacman/
Package manager for the grid
Saul Youssef (Boston)
Used by VDT
Single tool to easily manage installation and environment
fetch, install, configure, add to login environment, update
Sits over top of many software packaging approaches (rpm, tar.gz, etc.)
Uses dependency hierarchy, so one command can drive the installation of a
complete environment of many packages
Packages organized into caches hosted at various sites
Distribute responsibility for support
June 20, 2002Rob Gardner US Grid Projects 7
Grappa Workhttp://iuatlas.physics.indiana.edu/grappa/
Grid user interface for Athena
Flexible, use existing portal technology
Submit Athena jobs to grid computing elements
Manage JobOptions, record sessions
File staging and output collection supported
Registers output to MAGDA file catalog
Packaged for general ATLAS use with Pacman
Demonstrated on US ATLAS Grid Testbed (ATLFAST)
Future: virtual data portal + GANGA (UK Grid) collaboration
June 20, 2002Rob Gardner US Grid Projects 8
BU Tier 2
IU Tier 2
IU Physics clusterUniversity of Oklahoma
BNL
Athena Grid Job Submission
June 20, 2002Rob Gardner US Grid Projects 9
June 20, 2002Rob Gardner US Grid Projects 10
June 20, 2002Rob Gardner US Grid Projects 11
June 20, 2002Rob Gardner US Grid Projects 12
iVDGL in ATLAShttp://www.usatlas.bnl.gov/computing/grid/ivdgl/
Main goals for ATLAS
Integrate the GriPhyN VDT with ATLAS Core Software
Develop two prototype Tier 2 Centers for the US ATLAS
Integrate Tier 2 centers with US ATLAS Tier 1 Facility
Support development of US ATLAS Grid Testbed
Develop, integrate the US ATLAS piece of the iVDGL Laboratory
Year 1
Distribute, support VDT 1.1.x to US ATLAS sites
Tier2: 80K per Center for cluster upgrades and disk for ATLAS Data
Challenges
June 20, 2002Rob Gardner US Grid Projects 13
Indiana Tier 2http://tier2.iu.edu/
June 20, 2002Rob Gardner US Grid Projects 14
Boston Tier 2
June 20, 2002Rob Gardner US Grid Projects 15
Tier 2 Monitoringhttp://atlas.uits.iupui.edu/ganglia/index.php
June 20, 2002Rob Gardner US Grid Projects 16
Virtualize ATLAS Production
Track event data histories
Virtual data:
Transformation
Executable program
Athena service
Derivation
Execution of a transformation
Data Object
Named entity consumed by derivation
LFN, strings
Future: relational tables, persistent objects
Generator
HEPMC
Hits
Pileup Gen
Merged Hits
Simulation
Combiner
p-Hits
Digitizer
ROD Input
ROD emulator
Raw dataObjects
Reconstruction
Event Data
Objects
pythia.exe
atlsim.exe
Athena
services
Input cards
June 20, 2002Rob Gardner US Grid Projects 17
iSite
MAGDA-GDMP
Virtual Data Catalog
Virtual Data Portal
: Browser
: Executor
: Follower
Abstract Planner
Collection Catalog Replica Catalog
Concrete Planner
Replica Mgmt.
Services
Executor
:DAGMan
VDI
VDL
SEPersistency
Service
Ath
en
a G
rid
Exe
cutio
n W
rap
pe
rs
Algorithm Mgr.Services
Histogram,Monitoring, …
Services
Event SelectorServices
Globus GRAM, GridFTP, …
Co
re A
the
na
Scr
iptin
g S
erv
ice
s
Co
nd
or
Po
ol
Monitoring
CAS, SAS
Policy
Athena
Virtual Data System
June 20, 2002Rob Gardner US Grid Projects 18
Virtual Data Tools
Virtual Data API
A Java class hierarchy to represent transformations and derivations
Virtual Data Language
Textual for illustrative examples
XML for machine-to-machine interfaces
Virtual Data Database
Makes the objects of a virtual data definition persistent
Virtual Data Service
Provides an OGSA interface to persistent objects
June 20, 2002Rob Gardner US Grid Projects 19
VDT Status
VDT 1.1.0 equivalent installed on most sites
VDT 1.1.2 to be released after July 4
Globus 2.0
Condor 6.4.0
Condor-G 6.3.2
GDMP 3.0
ClassAds 0.9.2
Support has been set up