Cisco Demos HPC Technology
By CXOtoday Staff
Mumbai, Nov 17, 2006

Cisco will demonstrate its High Performance Computing (HPC) server interconnect technology at the SC06 International Conference this weekend. The family of server switching products offering availability, security with management tools, configuration interfaces, and a virtualization-provisioning platform designed to provide management regardless of the protocol used will be available in India, said company sources.

Vernon Turner, group vice president and general manager, Enterprise Computing practice, IDC said, "We are seeing continued adoption of highly distributed architectures to support compute intensive applications, not only within the HPC market, but within certain segments of enterprise data centers as well."

Cisco SFS Subnet Manager is designed for HPC environments where rapid cluster start-up and comprehensive diagnostic capabilities are critical. It is proven to scale beyond 4000 server nodes, has full database synchronization with redundant subnet managers, and is able to bring a multi-thousand-node InfiniBand cluster fabric up in less than one minute. For virtualization and dynamic resource provisioning, the company offers VFrame for InfiniBand fabrics.

"InfiniBand switches play an important role in our compute model and helped us achieve very scalable results with computational fluid dynamics and other simulation codes," says Steve Jones, founder of the Stanford University HPC Center and HPC manager for flow physics and computational engineering.

Cisco sources informed that the entire range of Server Fabric Switches for HPC and Grid computing environments are available now. Based on Infiniband technology, the server fabric switches range starts from $11,500 for a fixed 24-port configuration all the way to $355,000 for a 288-port Modular SFS switch.