I wrote a few weeks ago about the arithmetic behind server consolidation, and the number that keeps recurring in every conversation I have about it is the same one: a machine running at eight percent utilization still draws most of the power of a busy one, and virtualization is how you stop paying for capacity you are not using. That argument has not changed. What has changed since then is who is selling the tool that does it.
VMware has owned this market for the better part of a decade, and priced accordingly. ESX licenses are not cheap, and until recently that was fine, because VMware was the only mature option and enterprises paying six figures for hardware were not going to blink at a licensing line item that saved them from buying more of it.
That is about to get more complicated. Microsoft is shipping Hyper-V as a role inside Windows Server 2008, essentially bundled rather than sold separately, and while the current version is missing features VMware has had for years, live migration chief among them, it is going to be good enough for a large chunk of the market within a couple of releases. Citrix bought XenSource last year and is doing something similar from the open-source side.
None of this threatens VMware’s technology lead this year. It threatens the pricing model that lead has been sitting on top of.
What I think actually happens is VMware stops being able to charge like the only option in the room, because it is not going to be one anymore, and the company has to compete on management tools and support rather than on the hypervisor itself, which is quietly turning into a commodity the way most infrastructure eventually does.
The data centers actually doing the consolidating right now do not care much about any of this platform politics. They care about getting twenty racks down to three and getting their power bill under a number the utility company set for them, and whichever hypervisor gets them there fastest is the one winning this year regardless of what it costs to license.