Speaker
Dr
Simone Campana
(CERN/IT/PSS and INFN/CNAF)
Description
In the last 2 years, the ATLAS collaboration has been continuously running Monte
Carlo event simulation on the LCG/EGEE production grid. The ATLAS production system
is highly automated: jobs defined in a central database are handled by supervising
agents, submitted to and executed by the underlying Grid infrastructure. On the other
side, more recently the ATLAS Distributed Dada Management System also became fully
functional on EGEE. Such a system guarantees organization of data in a hierarchical
structure (datasets) and data distribution across sites via asynchronous file
transfer mechanisms. The ATLAS Production and the Distributed Data Management systems
have been now fully integrated on the EGEE infrastructure. Running the new framework
should improve the activity organization (production tasks assigned to specific
groups of tiers) and data distribution and availability (datasets automatically
aggregated in well defined computer centers). Data aggregation will be particularly
important for subsequent activities such as data analysis. This contribution provides
an architectural overview of the integrated system and a description of our
operational experience, showing the improvements in respect of the past, current
issues and possible future developments.
Author
Dr
Simone Campana
(CERN/IT/PSS and INFN/CNAF)
Co-authors
Dr
David Rebatto
(INFN MILANO)
Dr
Rodney Walker
(TRIUMF)