One of the fastest growing technology market segments is High Performance Computing (HPC), thus it should come as no surprise that Microsoft has tossed their hat into the ring. This long time Unix / Linux advocate has a hard time believing the Microbloated operating system now repackaged as Windows Compute Cluster Server 2003 or Windows Server 2003, HPC Edition depending on who you are talking to has been so streamlined that it will offer a competitive alternative to Linux in the HPC space.
The product is meant to compete in environment where customers are running scalable, parallel computing workloads in vertical market segments, such as engineering, life sciences and finance. Today this market is dominated by Linux. While some customer may look to Microsoft for turnkey HPC solutions it there are may unanswered questions.
Today many large HPC environments are leveraging clustered file-systems like Lustre (http://www.lustre.org/) or RedHat GFS formerly Sistina GFS (http://www.lustre.org/) which provide native fibre channel access to storage from the compute cluster. Some customers are using commercial offerings from companies like IBRIX (http://www.ibrix.com/) or Panasas (http://www.panasas.com/) who leverage NFS or proprietary clients to provide compute cluster to storage connectivity. These commercial vendors all build their code on Linux and target the HPC space, they are unable to drive CIFS performance to where it needs to be to satisfy the HPC market. Will Microsoft support these third party clients? Will Microsoft develop their own HPC storage server? Does Microsoft not realize that storage I/O is a critical component of a HPC environment?
I guess the question I have is…. Is the HPC market interested in point-and-click ease of use or raw speed? Maybe it’s me but when I walk into a shop filled with computational physicists I assume they enjoy a good challenge and that raw speed and performance are probably more important than a nice GUI.
One would have thought Microsoft would have learned that Linux can compete in niche spaces, but then again they continually loose the web server wars but they keep on trying. According to Netcraft’s June 2006 Web server survey the Apache, and thus Linux web server market share was 61.25 percent. The web server market place is another area where customers are interested in compute power and cost and ease-of-use and GUI seem to take a back seat. It will be interesting to see how this turns out.
Very interested in your thoughts on this one.
-RJB