Third Fastest Supercomputer in Full Production with World's Fastest Open Source Parallel File System, Lustre, from Cluster File Systems, Inc

Business Wire, Nov 14, 2003

Business Editors/High-Tech Writers

LIVERMORE, Calif.--(BUSINESS WIRE)--Nov. 14, 2003

The third fastest supercomputer in the world according to the 21st TOP500 supercomputing list (www.top500.org) at Lawrence Livermore National Laboratory (LLNL) has been in General Availability Production since October 1, 2003.

The 1,152 node, 11.2 teraflops peak, Multiprogrammatic Capability Cluster (MCR) integrates LLNL's Cluster High Availability Operating System (CHAOS) and Cluster File Systems, Inc.'s Lustre(TM) File System to provide a powerful research platform, with some 20 million files and 115.2 trillion bytes (terabytes) of aggregate storage across 64 object storage targets, that has already made significant contributions to LLNL research in the fields of laser research, fluid dynamics, materials behavior and seismology. MCR achieves 7.643 teraflops per second sustained as measured by the Linpack benchmark and with 90% efficiency for capacity production jobs is considered the fastest Linux-based supercomputer in the world.

Deployed with CHAOS, the Lustre File System has achieved I/O performance in excess of 4 Gigabytes per second.

"We started science runs several months ago to leverage MCR's power not only for the science but also to achieve levels of system and I/O performance that we have never before been able to offer," said Brian Carnes, Division Leader, Services and Development. "The science run opportunity achieves results for our users and early return on investment by stabilizing the hardware, perfecting CHAOS and Lustre and optimizing scheduling and operation of the cluster."

"In CHAOS we add to or modify tools and capabilities that allow Linux to achieve this high level of performance, reliability and robustness for our scientific applications," said Robin Goldstone, Group Leader, Production Linux, LLNL.

"Our work with the LLNL team led by Terry Heidelberg (Deputy Leader of High-Performance Systems) has significantly accelerated the completeness and maturity of Lustre as a high-performance and highly scalable solution," said Dr. Peter J. Braam, President and Chief Technology Officer of Cluster File Systems, Inc. "We look forward to pushing the performance envelope even further to meet the demanding performance and scalability requirements that are key to achieving LLNL's mission today and into the future."

About Lawrence Livermore National Laboratory

Founded in 1952, Lawrence Livermore National Laboratory is a national security laboratory, with a mission to ensure national security and apply science and technology to the important issues of our time. Lawrence Livermore National Laboratory is managed by the University of California for the U.S. Department of Energy's National Nuclear Security Administration. Laboratory news releases and photos are available at http://www.llnl.gov/PAO and on the UC Newswire.

About Cluster File Systems, Inc.

Cluster File Systems, Inc. (CFS) has quickly established itself as the recognized leader in high-performance, scalable cluster file system technology. Extensive experience, innovative insights, and proven engineering have enabled CFS to dramatically surpass the scalability limits of modern computing. The company's premier object-based clustered file system -- Lustre(TM) -- currently powers clusters with thousands of nodes and hundreds of terabytes of data, delivering groundbreaking parallel I/O and metadata throughput on some of the world's largest Linux-based supercomputers. CFS provides technical support, training, and engineering services for Lustre and is actively working with storage and cluster vendors to develop the next generation of intelligent storage devices. Lustre is Open Source software developed and maintained by CFS under the GNU General Public License.

Fore more information, see http://www.clusterfs.com