News

Clusters of plenty

• By Jack Vaughan
• June 9, 2004

Linux's maturing clustering and failover capabilities have quickly poised it as an attractive alternative to high-maintenance Unix systems, be they monolithic or distributed. In fact, Linux is shaking up the Unix world in a big way. But, Windows, too, is affected.

Young Linux clusters may not be bought for the traditional reasons -- in other words, high availability may not be as important a factor in their selection as is performance and cost. Cost cuts have been obtained because Linux is cheaper than Unix, and because Linux efforts are so much concentrated on the Intel chip architecture.

While it has been mainly an Intel-architecture phenomena, Windows clustering, still in its infancy, is also threatened by the march of Linux clusters. Both platforms look to supplant Unix. Applications that require high-server throughput or server consolidation are targets.

Among prominent startups working in clustering software is PolyServe. We spoke not too long ago with Steve Norall about PolyServe's doings in Linux clustering, and he gave a peek into some of the company's Windows plans, which were displayed in more detail at the recent Microsoft TechEd event. There, PolyServe released a version of its Matrix Server shared-data clustering system for the Windows platform. It also announced an alliance with Microsoft that places PolyServer as a Microsoft Certified Partner.

"Linux is being viewed as the replacement for the status quo Unix installed base that is out there. You have a value proposition that is extremely favorable, and that is to leverage the cost efficiencies of Intel-based software," said Norall, general manager, PolyServe Linux Solutions.

Traditionally, clustering has been for high-availability only. But companies like PolyServer have been pushing Linux -- and now Windows -- clustering for storage, performance and scalability.

Some advanced techies have been agog of late about grid, utility or on-demand computing. [Similar but different technologies seem to go through sets of similar but different names.] More clustering -- admittedly, this is a rather vague term -- is probably what is really going to occur.

PolyServe's Norall says his software does not require developers to change the way to do things. The infrastructure software, he implies, should take care of the deployment details. But he does suggest that how you design and program software will change if you know the application you are building can be "parallelized." Much of the interest in Web services revolves around just such a premise.