IBM is looking into ways it can leverage its Blue Gene supercomputer within the company's BladeCenter line of ultra-thin blade servers, a company executive has confirmed.
"We've been looking into ways to commercialize what we have with Blue Gene, and using that technology within BladeCenter is something that would make sense for us to do," said Jeff Benck, vice president and business line executive for IBM's eServer BladeCenter. Benck was in London on Friday demonstrating IBM's blade server technology to local press.
According to Benck, nothing concrete will come from blending Blue Gene technology with blade servers this year, rather a 2006-2007 time frame would be more likely.
IBM, in Armonk, New York, has been working toward building a commercial version of the Blue Gene supercomputer, called Blue Gene/L, aimed at the enterprise as well as the scientific and research markets.
Lawrence Livermore National Laboratory in California has a Blue Gene/L, which is ranked as the world's fastest supercomputer, featuring 32,000 two-processor nodes packed into a compact area. The computer is currently being extended and by June is expected to grow into a 130,000-processor system featuring a peak performance of 360 teraflops, all fitting into 64 server racks.
In March, IBM began offering Blue Gene supercomputing on demand, allowing customers to run Linux-based workloads through a dedicated VPN (virtual private network) into its Deep Computing Capacity on Demand Center.
According to IDC analyst Dan Fleischer, the combination of Blue Gene technology with the BladeCenter line is a natural evolution for IBM as it seeks to make its on demand utility computing model more appealing to enterprise customers.
"Blue Gene is more about pushing the utility computing concept, and blade servers make a natural fit to employ Blue Gene technology on," Fleischer said. "Blades are the form factor of choice."
But IBM faces an up-hill battle when it comes to winning over the enterprise market, Fleischer said, in part because utility computing is a very complex concept to deploy.
"Enterprises would like to use utility-type models, but issues like legacy infrastructures mean they aren't in a position to deploy the technology. Enterprises still aren't sure how to roll out utility computing within an organization and getting there is something they're grappling with," Fleischer said.