Failing the 'Sniff' Test
This IDC report
on Microsoft's huge increase in hypervisor share, up to 23 percent of the market, has been widely covered in news reports and blogs. And it's easy to see why. From the IDC press release:
"When looking at new x86 virtualization licenses, VMware continues to hold on to its strong position in the x86 market with a combined market share of VMware ESX and VMware Server at 44%. However, in its first quarter of general availability Microsoft Hyper-V delivered a strong showing, and when combined with Virtual Server 2005, Microsoft's market share is 23% of new shipments."
Those are significant figures, enough to make any journalist or blogger sit up and take notice. There's only one problem with them:
Sometimes you have to apply common sense, even to analyst reports from a company as respected as IDC. And common sense says to this virtualization watcher that those figures must be way, way off. Does anyone who follows this industry really believe that Microsoft is already at half of VMware's market share for new hypervisor shipments? Good grief, Hyper-V's only been available for four months. Are we supposed to believe that it, in concert with Microsoft Virtual Server (a insignificant product in terms of enterprise virtualization usage), Redmond has captured a quarter of the market?
As Al Borland used to say to the Tool Man, "I don't think so, Tim."
So if they are bogus, what accounts for such Fantasy Island numbers? It's hard to know exactly, but Mike DiPetrillo of VMware has hazarded some theories. Keeping in mind that Mike is less than an objective observer, I think he makes a number of very good points.
In addition, the IDC figures clash substantially with those of Gartner, as Alessandro points out over at virtualization.info. Somebody's wrong somewhere, but the Gartner figures jibe more closely to what my common sense tells me. I discussed this with Virtualization Review columnist, and well-respected Burton Group analyst, Chris Wolf this morning, and he was similarly suspicious of the IDC numbers. Of course, Burton Group is an IDC competitor; it's worth noting, however, that Burton does not do any vendor-sponsored reports, the way IDC and Gartner do. Chris has nothing to gain or lose by stating his opinion in this.
I'm also not claiming, as Mike D. strongly implies, that the IDC study was commissioned by Microsoft. Although it may have been, I have no evidence of that. If the IDC study was commissioned, they absolutely should say so, and by whom (Side note: the relationship between many analyst firms and vendors is far too incestuous. I'm more than skeptical of any sponsored report, study or survey. I'm downright cynical, in fact, to the point of "guilty until proven innocent." I assume there's a bias at work that influences the results, unless I'm convinced otherwise. Back to your regularly scheduled programming.)
So, what are my conclusions on this report? It's a bunch of hokum. Much more serious analysis needs to be done before this kind of information should be put out there. In other words, do your homework better, IDC.
I would also caution my fellow journalists to do more of a "sniff" test on these types of things. Does a purported fact make sense on its face? If it doesn't, you might wanna hold off and do some more reporting.
As a P.S., I have to disagree with one other quote in the IDC report. Analyst Brett Waldman says: "IDC believes that the high-volume consolidation opportunities - the low hanging fruit in the x86 server virtualization market -- is starting to dry up. This is, in turn, resulting in smaller deals overall." That quote doesn't square with the latest estimates that about 12 percent of servers are virtualized. That leaves quite a harvest of low-hanging fruit still on the trees. Doesn't sound like it's drying up to me -- even though server consolidation may be seen as passe these days.
Posted by Keith Ward on 10/31/2008 at 12:48 PM