Powers Unfiltered

Well, it looks as though the Linux community has noticed Microsoft’s entry into the HPC market. 

In reviewing Tom Groenfeldt’s financial services blog on MSDN, I ran across a comment that led back to a longer post by Brad Chamberlin, expressing his incredulity that anyone could take an HPC offering from Microsoft seriously.  Brad’s comments are typical of the Linux community’s response to Microsoft’s Windows Compute Cluster Server 2003 (CCS), and they are as predictable as they are one-sided. 

What a joke, how can you “take back” market share you never had in the first place.  

Besides that, $469/node vs. FREE????

Certainly, Brad is right in pointing out (pedantically) that Microsoft can’t “take back” market share that it never had.  As this is the first dedicated HPC product Microsoft has launched, Brad knows this is a bit of sloppy reporting by the original source he cites.  Brad provides an analysis of the respected Top500 list, and concludes:

So as you can see with this little jog through the top500 list Microsoft Windows market share in HPC is NILL, NOTHING, NONE, NON-EXISTANT, NADA. You can’t take back something you never had!

I am sure with it’s $469/node price tag Linux has a lot to be worried about. I wouldn’t be surprised if by 2007 Windows holds 2, maybe 3, whole spots on the top500 list.

But if Brad is so convinced of the future market share to be held by the Windows OS in HPC, then why choose the Top500 at all?  As a measure of market share, this is specious; the Top500 list, by definition, rounds to approximately zero percent of the market for high performance computing.  Also, it represents the zero percent that Microsoft specifically identified as outside its target market. 

With Windows Compute Cluster Server 2003, Microsoft is going after not the Top500, but the next 500,000.  Microsoft has been very open about this, saying they would bring HPC into the mainstream.  And if the HPC market in 2007 is still mostly Linux, well, they’ll be around in 2008 and 2009 and 2010 to keep bashing away. 

Brad’s point about Linux being “FREE” is also the same simplistic argument that’s been beaten to death — in most enterprises, a supported version of Linux is not free, and the OS cost is a tiny fraction of the total cost of ownership, so it’s really about value delivered, blah blah blah.  So the comment “$469/node vs. FREE????” adds nothing new — yes, Brad, at $469/node, there are definitely businesses that will want the Active Directory integration, ease of deployment, compatibility with a wide range of Windows software, and other nice features that Microsoft has put into CCS. 

The battlefield is littered with folks as complacent as Brad who underestimated Microsoft’s persistence, resources, and market savvy.  While CCS has some rough edges, it’s better than most “1.0” products, and it’s only going to get better.  When Microsoft enters a market, they don’t always dominate, but it certainly does not pay to be complacent.   Rather than dismiss CCS, let’s watch and see what Microsoft and its partners can do in this market.  Brad may be surprised.