The ATLAS experiment at the Large Hadron Collider (LHC) has begun the work in November, 2009 in the European Organization for Nuclear Research (CERN) in Geneva. Now the data are accumulated at the proton collision energy of 7 TeV. It is expected that the volume of the data from the ATLAS setup will make 3.5 PB/YEAR, at full design luminosity of the accelerator. Data processing includes reconstruction of events in the detector, statistical simulation and the analysis. According to model of data processing of the experiment, the use of geographically distributed computing resources is provided. For the organization of work of the computer centers of ATLAS VO, the three-level grid-infrastructure is used. Computer center of the JINR is a part of the Russian united ATLAS center of the 2nd level (RuTier-2). According to the role in this center, the JINR computing center should give computing resources for the analysis and statistical simulation, and also provide the storage of 1/5 of all experimental data of ATLAS (in the simplified format optimized for the analysis).
The ATLAS VO grid-infrastructure of the JINR computing center includes tools for installation of the software of experiment, data management system, and tools of statistical simulation job management and the distributed data analysis.
The ATLAS data management system (DDM) provides automatic copying of the data between the JINR and the SARA center of the first level (Amsterdam, the Netherlands), and also between the JINR and other centers of the second level located in Russia, Israel, Turkey and Ireland, and associated with Tier-1 SARA. Besides, as appropriate, the data management system allows transferring data to/from any other computing centers of the ATLAS VO.
Submission and management of ATLAS statistical simulation jobs in JINR is accomplished in a centralized way from the CERN. The state and result of job execution is constantly traced and published on the page of job management system of the experiment. Simulated data are transferred for storage to the Tier-1 SARA. Submission and management of the distributed analysis jobs is carried out by two auxiliary programs — GANGA and pathena.
For reliability control of work of the JINR computing center as a part of a computing infrastructure of ATLAS experiment, along with storage of experimental data, simulation and analysis, automatic checks of functionality of the JINR grid-segment are constantly carried out. They include tests of data transfer system, and stress-tests of system of the distributed analysis of the data.
Installation and management of versions of the software of experiment is carried out centrally from the CERN. All software, including tools of the distributed analysis and data control of ATLAS, is accessible to employees of the JINR – participants of the ATLAS experiment, on computers of the common access of the JINR computing center.
Since recently an important activity of setting up of ATLAS Tier-3 centers is rapidly evolving. At JINR the ATLAS Tier-3 center is created on the base of the computing cluster in the Laboratory of Nuclear Problems. It is aimed both to the data analysis and software development. At the same time, in the Laboratory of Information Technologies of JINR a virtual computing infrastracture has been set up, to facilitate development and testing of tools to monitor existing and future ATLAS Tier-3s.
ATLAS Computing technical interchange meeting (the official ATLAS Collaboration computing meeting) has been held at JINR on 31.05.2011 – 02.06.2011 (see the meeting program).
Training and help to physicists experts in the questions, concerning work in the grid environment, are prominent aspects of ATLAS VO support in the JINR. In 2007-2008 in the JINR the practical training on the distributed analysis of data of the ATLAS with the use of the grid has been provided. User support proceeds now.
More detailed information can be found on the ATLAS group web-page at JINR.