Hyping Hyper-V

Virtualization is becoming a core strategy for Microsoft. It sees an opportunity in a market that's still at less than 10 percent saturation, and is pouncing. If you need evidence, just look at how quickly it's been pushed through the pipeline: the first public beta was early, the first release candidate was early, and RTM is at least a month, and probably more, early (no definitive date was ever given for the final bits to be released, but spokesperson after spokesperson repeated the "180 days after release of Windows Server 2008" mantra).

It's interesting and ironic that Windows 2008 was one of the most delayed products in Microsoft's long and gory history of delays (initial Microsoft predictions pegged a 2005 release for then "Longhorn" server), but Hyper-V, one of its key pieces, has been slingshotted through the process. It's clear that Microsoft sees the iron as white-hot now, and wanted to strike before getting any further behind VMware's ESX (or Citrix' XenServer, for that matter).

(One interesting side note from my conversation yesterday with Microsoft's Arun Jayendran. I asked specifically if Hyper-V will have an embedded version, similar to ESXi or the embedded versions of XenServer offered by the competition, that hardware OEMs can ship out the door. Jayendran declined comment. That says to me that Microsoft is at least considering it; otherwise I would have expected a "No." Update I've been informed that "Hyper-V Server", the standalone version, will be available to OEMs, same as the other embedded hypervisors. No word on whether it will be tweaked for delivery that way.)

Remember that Hyper-V is an incomplete product. A number of key features, like Live Migration, hot-swap ability, and support for more processors was killed off way back when it was still code named "Viridian." Jayendran affirmed that at least one of those features, Live Migration, will be added with the next version of Hyper-V, likely to be released with the R2 release of Windows 2008.

Still, it's hard to think of a better "Version 1" of a Microsoft product -- ever. I've used it some, and have talked to tons of folks who have used it in beta and in production. They almost uniformly rave about it. In fact, I talked to a consultant who has a number of customers running it in production every day, and have been since the beta days. This consultant, who has no axe to grind -- he makes his money by keeping his clients happy, not by pushing one product over another because of "special deals" or vendor relationships -- now routinely recommends Hyper-V over ESX. And it's not that he hates ESX; quite the opposite. It's just that, with as well as Hyper-V works, he sees a better value proposition in most cases using that instead of VMware.

Does that scare you, VMware? It should. In fact, I expect to see a drastic ESX price cut announced before the end of the year. If VMware thinks it can laugh off Hyper-V, it is sadly -- perhaps fatally -- mistaken. The reality is that ESX is better than Hyper-V; better by a good bit, in fact. The question is whether it's so much better that it can continue to command a premium price. Hypervisors aren't a money-maker anymore; the competitive market, established today by Redmond, will quickly decide that, in my opinion. Heck, even the GUI-less ESXi will cost you $500 a pop, minimum. VMware will have to respond.

The battle has been joined. Will it be to the death?

Posted by Keith Ward on 06/26/2008 at 12:48 PM