headerlogo
scyourway
Award Finalist/Winner

SC Conference - Activity Details



Data-lifecycle Management over a Loosely Coupled Distributed Infrastructure for LSST's HPC Applications

Team Members:
Arun Jagatheesan  (iRODS.org / San Diego Supercomputer Center)
Michael Wan  (iRODS.org)
Wayne Schroeder  (iRODS.org)
Chris Cribbs  (National Center for Supercomputing Applications)
Ray Plante  (National Center for Supercomputing Applications)
Jeff Kantor  (Large Synoptic Survey Telescope)
Tim Axelrod  (Large Synoptic Survey Telescope)
Chris Smith  (National Optical Astronomy Observatory)
Petri Garagorri  (National Optical Astronomy Observatory)
Ron Lambert  (National Optical Astronomy Observatory)
David Walker  (National Optical Astronomy Observatory)
Mark Holliman  (University of Edinburgh)
Elena Breitmoser  (University of Edinburgh)
Robert G. Mann  (University of Edinburgh)
Arthur Trew  (University of Edinburgh)
Jean-Yves Nief  (Centre de Calcul IN2P3)
David Gehrig  (University of Illinois at Urbana-Champaign)
Reagan Moore  (University of North Carolina at Chapel Hill and Renaissance Computing Institute)
Arcot Rajasekar  (University of North Carolina at Chapel Hill and Renaissance Computing Institute)
Challenges Session
SC08 Storage Challenge
Tuesday,  10:30AM - 11:00AM
Room 17A/17B
Abstract:
The Large Synoptic Survey Telescope (LSST) will produce more than 15 terabytes of image data every night during its observation. Nature's editorial mentions LSST's data management as "Steering the future of computing". -------------------------------------------------- LSST infrastructure software that manages data storage has three distinct requirements. Firstly, it has to provide high performance I/O that will be required by the LSST middleware and the LSST applications. Secondly, the LSST infrastructure software has to effectively scale and manage the lifecycle of hundreds of millions of files and their replicas to satisfy requirements from both HPC scientific-applications and public dissemination of data to a diverse user community over a decade of operation of LSST. Thirdly, the infrastructure software has to ensure a highly flexible data storage infrastructure that can dynamically allow addition/removal of participating collaborators and their storage systems. -------------------------------------------------- We will demonstrate different aspects of our solution with respect to LSST's data storage lifecycle management.
   IEEE Computer Society  /  ACM     2 0   Y E A R S   -   U N L E A S H I N G   T H E   P O W E R   O F   H P C