Cargando…

Distributed Data Analysis in ATLAS

Data analysis using grid resources is one of the fundamental challenges to be addressed before the start of LHC data taking. The ATLAS detector will produce petabytes of data per year, and roughly one thousand users will need to run physics analyses on this data. Appropriate user interfaces and help...

Descripción completa

Detalles Bibliográficos
Autor principal: Nilsson, P
Lenguaje:eng
Publicado: 2012
Materias:
Acceso en línea:http://cds.cern.ch/record/1445597
_version_ 1780924773600067584
author Nilsson, P
author_facet Nilsson, P
author_sort Nilsson, P
collection CERN
description Data analysis using grid resources is one of the fundamental challenges to be addressed before the start of LHC data taking. The ATLAS detector will produce petabytes of data per year, and roughly one thousand users will need to run physics analyses on this data. Appropriate user interfaces and helper applications have been made available to ensure that the grid resources can be used without requiring expertise in grid technology. These tools enlarge the number of grid users from a few production administrators to potentially all participating physicists. ATLAS makes use of three grid infrastructures for the distributed analysis: the EGEE sites, the Open Science Grid, and NorduGrid. These grids are managed by the gLite workload management system, the PanDA workload management system, and ARC middleware; many sites can be accessed via both the gLite WMS and PanDA. Users can choose between two front-end tools to access the distributed resources. Ganga is a tool co-developed with LHCb to provide a common interface to the multitude of execution backends (local, batch, and grid). The PanDA workload management system provides a set of utilities called PanDA Client; with these tools users can easily submit Athena analysis jobs to the PanDA-managed resources. Distributed data is managed by Don Quixote 2, a system developed by ATLAS; DQ2 is used to replicate datasets according to the data distribution policies and maintains a central catalog of file locations. The operation of the grid resources is continually monitored by the GangaRobot functional testing system, and infrequent site stress tests are performed using the HammerCloud system. In addition, the DAST shift team is a group of power users who take shifts to provide distributed analysis user support; this team has effectively relieved the burden of support from the developers.
id cern-1445597
institution Organización Europea para la Investigación Nuclear
language eng
publishDate 2012
record_format invenio
spelling cern-14455972019-09-30T06:29:59Zhttp://cds.cern.ch/record/1445597engNilsson, PDistributed Data Analysis in ATLASDetectors and Experimental TechniquesData analysis using grid resources is one of the fundamental challenges to be addressed before the start of LHC data taking. The ATLAS detector will produce petabytes of data per year, and roughly one thousand users will need to run physics analyses on this data. Appropriate user interfaces and helper applications have been made available to ensure that the grid resources can be used without requiring expertise in grid technology. These tools enlarge the number of grid users from a few production administrators to potentially all participating physicists. ATLAS makes use of three grid infrastructures for the distributed analysis: the EGEE sites, the Open Science Grid, and NorduGrid. These grids are managed by the gLite workload management system, the PanDA workload management system, and ARC middleware; many sites can be accessed via both the gLite WMS and PanDA. Users can choose between two front-end tools to access the distributed resources. Ganga is a tool co-developed with LHCb to provide a common interface to the multitude of execution backends (local, batch, and grid). The PanDA workload management system provides a set of utilities called PanDA Client; with these tools users can easily submit Athena analysis jobs to the PanDA-managed resources. Distributed data is managed by Don Quixote 2, a system developed by ATLAS; DQ2 is used to replicate datasets according to the data distribution policies and maintains a central catalog of file locations. The operation of the grid resources is continually monitored by the GangaRobot functional testing system, and infrequent site stress tests are performed using the HammerCloud system. In addition, the DAST shift team is a group of power users who take shifts to provide distributed analysis user support; this team has effectively relieved the burden of support from the developers.ATL-SOFT-SLIDE-2012-155oai:cds.cern.ch:14455972012-05-03
spellingShingle Detectors and Experimental Techniques
Nilsson, P
Distributed Data Analysis in ATLAS
title Distributed Data Analysis in ATLAS
title_full Distributed Data Analysis in ATLAS
title_fullStr Distributed Data Analysis in ATLAS
title_full_unstemmed Distributed Data Analysis in ATLAS
title_short Distributed Data Analysis in ATLAS
title_sort distributed data analysis in atlas
topic Detectors and Experimental Techniques
url http://cds.cern.ch/record/1445597
work_keys_str_mv AT nilssonp distributeddataanalysisinatlas