US CMS and OSG Plans for 2014

~Burt Holzman

With the New Year behind us, I would like to take a look ahead at what we have planned for 2014. LHC beam does not return until 2015, but we have plenty to do in the meantime!

US CMS plans to expand the use of multicore jobs, leveraging glideinWMS (the glidein-based Workload Management System) and HTCondor to allow us to dynamically partition resources as late as possible. Additionally, we will begin channeling both analysis and production jobs into a unified pool, allowing the experiment to change priorities easily and on the fly. We will expand our use of our global xrootd federation, enabling efficient data access over the WAN and a unified namespace — users should be able to access their data without needing to know where the disks are physically located. We are also going to increase our use of opportunistic resources, from OpenStack clouds available at CERN and elsewhere to grid resources from the OSG. This increased compute capacity naturally needs to read increased amounts of data, and our Tier 1 storage facilities are being reconfigured to allow centralized control over what data is resident on disk to increase the overall processing efficiency. With all of this work, we also are planning detailed tests to exercise the full production and analysis chain, and we look forward to leveraging the tools and facilities of OSG to get us ready for data taking in 2015!