Windows Server 2003 Compute Cluster Edition

Over the past few weeks, we discussed Microsoft's Automated Deployment Services (ADS). Last week, we discussed patch management in a Microsoft world. This week, we look at Microsoft's latest attempt to take on Linux.

At a recent conference, Microsoft demonstrated its latest addition to the Windows Server 2003 operating system that will target Linux and Unix-based grid computing and supercomputers. The new OS, called Windows Server 2003 Compute Cluster Edition (previously, the company had referred to the product as Windows Server HPC Edition), will attempt to gain a foothold in the growing grid computing market space, in an effort to gain some dominance over the popular Linux and Unix systems that currently make up the majority of this space.

What is grid computing? In its most basic sense, grid computing is the sharing of multiple processors on multiple machines in an effort to increase the compute horsepower for CPU-intensive applications. Grid computing can be setup in a controlled manner in which you build multiple servers, running similar operating systems, and similar grid software. Alternatively, you can take advantage of resources (i.e. CPUs) on many disparate systems throughout your enterprise by installing the appropriate grid software on each of the servers/workstations within your company. Then, by utilizing a centralized queue management system, you would be allowed to submit jobs to be computed across many processors concurrently - thus enabling an application to take advantage of tens, or even thousands, of processors to compute the results of a compute-intensive application.

With the release of Windows Server 2003 CCE, Microsoft could potentially leverage its strong PC install base as a way for compute intensive applications to share processors across all of the Windows systems within your company. This could be a major impact to Linux, as well as some of the popular grid computing software that exists today.