May 12, 2005

SDSC Boosts Data Capabilities
Center now provides 1.1 Petabytes of on-line disk storage

By Ashley Wood

The San Diego Supercomputer Center at UCSD recently added new on-line disk storage, offering 1.1 petabytes to users, complementing its 6+ petabytes of tape storage, 4.2 terabytes of DataStar supercomputer memory, and other data and storage capabilities. The addition will boost the resources provided to science and engineering users and will better support use of the over 70 community databases and data collections housed at SDSC.

The expansion IBM Serial-ATA online disk capacity to SDSC’s Storage Area Network will make it one of the largest data storage resources in the nation for academic research. One petabyte of stored data would fill the Library of Congress more than eight times over.

SDSC’s leadership focus is data-oriented science and engineering applications which not only have high-capacity storage requirements, but also require high bandwidth for moving data from compute systems to/from disk, and flexible access across distributed grids. The center’s new disk storage will be leveraged by these applications: The national TeraGrid project will utilize the storage as a large disk cache to support a Global File System (GFS) across the TeraGrid distributed grid. It will also provide high-speed storage for SDSC’s data-oriented BlueGene system. It will also become a core component of SDSC’s integrated system to host a diverse set of shared science, engineering, and societal data collections, which can be used in connection with SDSC’s computing resources or accessed across a network.

"The additional storage will help us do an even better job of providing a world-class environment to users for data management, data-oriented computing, and data cyberinfrastructure,” said Dr. Fran Berman, Director of SDSC and holder of the Endowed Chair in High Performance Computing at UCSD's Jacobs School of Engineering. "We are excited about being able to offer these enhanced capabilities and capacity to the broad science and engineering community.”

SDSC has also completed a significant upgrade to its HPSS (High-Performance Storage System) archival system which included more disk cache, improved networking and a software update to HPSS 5.1. HPSS was re-hosted on two of DataStar’s p690 nodes. In addition to improved availability, these updates should enable users to see increases in the read/write rates of two to eight times depending on the applications.

Communities will able to immediately utilize this enhanced capability. “We’re currently working with the National Virtual Observatory (NVO) to put their all-sky survey data on disk for access across the TeraGrid," said Anke Kamrath, Director of User Services at SDSC. “By being able to "mine" this entire data collection, NVO astronomers will be able to reduce the time it takes discover hard-to-find objects such as brown dwarf stars.” The storage resource will enable researchers to answer big questions in many areas, from simulating large earthquakes and deciphering the tree of life to exploring the events of the early universe.

“These facilities are critical for leadership data cyberinfrastructure” said Berman, “and critical to help the science and engineering community achieve the next generation of advances and discoveries.”

Media contacts:
Greg Lund, SDSC Communications, (858) 534-8314 or
Ashley Wood, SDSC Communications, (858) 534-8363