INDUSTRY

CFS & Partners Provide Storage Solutions Delivering Production-Caliber Uptime -- Cluster File Systems, Inc. (CFS), the leader in high-performance parallel file systems, is celebrating the third year of general availability of its open-source Lustre technology by reporting widespread file system stability and robust customer implementations on five continents. The Lustre file system is a next-generation cluster storage solution, designed to serve clusters with up to tens of thousands of nodes, manage petabytes of storage, and transfer hundreds of gigabytes per second with state-of-the-art security and management infrastructure. Over the past 2 years, Lustre technology has been deployed on many of the world's largest cluster computers, and is now rapidly gaining global acceptance in both government and commercial high-performance cluster environments. Today, users affirm that the Lustre file system provides the stability and robust architecture required to support critical research and business objectives. At the National Center for Supercomputing Applications (NCSA), Lustre software is used on the 1,280-node Tungsten cluster manufactured by Dell. The Lustre file system consists of 104 Object Storage Servers and 140 terabytes of DataDirect Networks storage. "The Lustre File System on our Tungsten cluster is very stable -- with the cluster delivering 99% uptime consistently. This system is routinely over-requested by the National Science Foundation users and Tungsten is regarded as a very stable resource. The high quality of our Lustre deployment is critical in helping us achieve this stability," says Michelle Butler, Technical Program Manager at NCSA. "We are routinely achieving greater than 90% utilization on our Lustre-based system. The performance, stability and support of the Lustre file system can be relied upon in our computational environment," said Tom Kendall, Lead Systems Engineer at the Army Research Laboratory Major Shared Research Center, whose Lustre file system supports the 1,024-node John Von Neumann cluster delivered by Linux Networx. "Cray and CFS have worked closely together to increase Lustre reliability for very large-scale systems. We have shipped our industry leading Cray XT3 and Cray XD1 supercomputers around the world with the Lustre file system, with configurations of over 10,000 processors. These systems have passed customer acceptance tests and are currently being deployed for production-level computing at scale," said Jan Silverman, Cray Senior Vice President, Corporate Strategy and Business Development. In Sweden, a Lustre system helps accelerate drug discovery on a high-performance Linux cluster at AstraZeneca. "Our evaluation has very rapidly moved into production mode, thanks to a really stable product," says Mathias Gustavsson, Linux System Manager at AstraZeneca R&D. Within CFS, important processes are followed to ensure that further innovation doesn't come at the expense of reliability or performance. "The continuing stabilization of our product has benefited tremendously from our partnership with Carnegie Mellon's Software Engineering Institute," Phil Schwan, CFS's Chief Executive, said. "And with the support of our other vendor partners, we ensure an extremely high level of product quality and hardware integration. These positive customer experiences tell us that we're making the right choices."