出版社:SISSA, Scuola Internazionale Superiore di Studi Avanzati
摘要:Higher instantaneous luminosity of the Tevatron Collider and improved detector and trigger efficiencies
require large increases in computing power for the CDF experiment in order to be able to
meet future needs of data analysis and MC production. So far, CDF has been relying on a set of
Analysis Farms (CAFs) with dedicated resources of commodity nodes managed as Condor pools,
with a small CDF specific software layer on top of Condor. However, it has long been understood
that significant expansion of the dedicated pools will be impossible in the future, and that CDF
must move towards utilizing shared Grid resources. We have extended the existing CDF computing
model by incorporating the Condor glide-in mechanism that can create dynamic Condor
pools on top of existing batch systems, without requiring any additional software installation. All
the monitoring tools supported on the dedicated CAFs, including semi-interactive access to the
running jobs and detailed monitoring, have been preserved. CDF is also building an interface over
the gLite tools for a broader access of resources and looking into other possibilities. In this paper,
we will discuss about various CDF Grid initiatives and how they fit into the CDF plans.