Fabric Networks to deliver 128 node InfiniBand supercomputer cluster to Sandia National Labs
Largest fully integrated InfiniBand cluster in the World

WESTBOROUGH, MA - June 30, 2003 - Fabric Networks, Inc., the leader in switched fabric networking technology, today announced that they have received an order from Sandia National Labs for the installation of a 128-node Linux cluster. The cluster will use InfiniBand switched fabric networking technology from Fabric Networks. This will be the first public, fully integrated cluster of this magnitude in the United States based on 10 Gb/s InfiniBand technology. The cluster will be used to test the performance of scientific computing applications and InfiniBand management features at scales not previously possible.

"What interests us about InfiniBand is the combination of low-latency, high-bandwidth, low CPU utilization, and open standards," said Curt Janssen, principal member of the technical staff at Sandia National Laboratories. "We expect InfiniBand to continue to track advances in processors' local interconnect technology and thus provide a foundation for clusters now and in the future."

"Recently, we tested the scalability and performance of InfiniBand at our benchmarking facility, demonstrating to Sandia our capability to deliver InfiniBand-based cluster systems," said Joshua Harr, chief technical officer of Linux Networx. "Linux Networx has also validated Fabric Networks' InfiniBand products on our Evolocity® cluster systems, which allowed us to make this joint customer win."

InfiniBand is a new network technology based on an open standard I/O architecture developed by a consortium of some of the world's largest computer companies, including Intel, Hewlett-Packard, IBM, Sun, and Microsoft. InfiniBand offers data rates 10 times higher and latencies 10 times lower than that of Gigabit Ethernet. It also reduces communications overhead by as much as 50% of server CPU cycles.

Linux Networx will supply the computing platform, cluster management tools, storage, and parallel programming libraries for the cluster. The cluster consists of 128 Evolocity II high-density nodes, each with GHz Xeon® CPUs. Each node will have a Fabric Networks HCA-X card connected to the FabNet™ 12800 (twelve eight-hundred), a 128 port InfiniBand switch. The cluster will be running Redhat Linux. Clusterworx? software from Linux Networx will manage cluster resources and applications while FabNet Fabric Manager will manage the switched fabric.


About Fabric Networks, Inc.

Fabric Networks, Inc. is the leading provider of switched fabric networking solutions for enterprise data center high availability computing and High Performance Computing server clusters. Created from the strategic merger of InfiniSwitch Corporation and Lane15 Software, Fabric Networks offers a legacy of excellence in both switched fabric hardware and fabric management software. Designed to improve CPU utilization, cluster scalability and latency, Fabric Networks fully integrated solutions are standards-based and fully interoperable with InfiniBand 1.1 compliant products. Fabric Networks solutions offer outstanding performance, high availability, plug-and-play scalability and smooth integration with existing environments. Fabric Networks solutions have attained IBM ServerProven® status and an OEM partner with other leading server manufacturers. For more information, visit www.FabricNetworks.com.

InfiniBand is a registered trademark of the InfiniBand Trade Association. ServerProven is a registered trademark of IBM Corporation. IBM makes no guarantee of the compatibility of specific products. Fabric Networks and FabNet are trademarks of Fabric Networks, Inc.






© 2008 Flagship Ventures. All rights reserved.