In the interest of helping to ensure software continuity for customers, IBM Wednesday married Linux clusters and blade servers to offer a pre-packaged "supercluster." The idea is to pack as much power as possible in as small a space to improve system manageability, a challenge for blade servers.To do this, Armonk, New York's IBM has taken the capabilities of its existing eServer Cluster 1350 machine and software and bundled it with its eServer BladeCenter to make installing and managing Linux clusters a bit more efficient.
The eServer Cluster 1350 can be built with any combination of IBM eServer BladeCenter systems, such as the IBM eServer x335 and eServer x345 units. The cluster can also combine eServer x345 or eServer x360 storage nodes, as well as an eServer x345 management node, using Intel Xeon processors running at speeds up to 3.06 GHz. The cluster management software for the Cluster 1350 is the Linux version of the same software used in IBM's Deep Blue supercomputer.Cluster computing draws on multiple computers, such as workstations, multiple storage devices, and redundant interconnections, to form what appears as one powerful system. A common use of cluster computing is to load balance traffic on high-traffic Web sites. The concept of high-performance technical computing (HPTC) has been gaining momentum since the '90s, and vendors such as IBM, Hewlett-Packard and Dell Computer have been increasingly applying clustering capabilities to everyday IT solutions to help enterprises run more smoothly.Blade servers are a newer trend, albeit one adopted by the same systems vendors who work in the HPTC space. Blades pack the power of larger boxes into thin servers, which are inserted into a chassis and managed by software to cut back on much of the cables and hookups associated with refrigerator-sized machines. In this case, IBM will add eServer Cluster 1350 systems to its BladeCenter to save space.
Whereas HPTC computing of the past mostly focused on life sciences or climate prediction and tracking, IBM Vice President of Deep Computing Dave Turek told internetnews.comthe bolstered Cluster 1350 is a building block that will "provide the power behind everything from petroleum exploration to digital animation."
Turek said the revised machine would appeal to all segments of the market because customers no longer want the hassle of crafting their own systems.
Sageza Group Research Director Charles King agreed, noting that when firms wanted clustered supercomputing, they had no place to turn to. Simply, IBM has taken the idea of a cluster and blade server and commoditized it.
"Basically, IBM has created an off-the-rack commercial product that customers can buy and create a plug-in for it," King told internetnews.com. "In the past when vendors put a high-performance technical computing-oriented machine together, it would often be a one-off, custom built machine, or came from pieces built by a research lab."
Turek said the Cluster 1350 upgrade is just a taste of a slew of announcements IBM plans on issuing over the next several months from its Deep Computing division, which was formally namedlast month. He wouldn't divulge specifics, but said eServer Cluster 1350 is now a building block to additional software continuity projects and products that company is developing to address systems management.
The eServer Cluster 1350 supports IBM's FAStT200 and FAStT700 storage and IBM has integrated an optional fibre switch into the BladeCenter chassis. The Cluster 1350 with the IBM eServer BladeCenter will be ready for the public June 6. Pricing has not been made public, but IBM indicated it would be on a sliding scale, based on cluster configuration requirements.