Betting on virtualization

I am not a betting man (I paid attention in class when we covered probability) but I would bet that virtualization will change everything.

I concluded last week's column with the prediction that virtualization would make the server business more cutthroat than it already is.

My reasoning was that when hardware is isolated from the operating systems to the extent that it is effectively generic and any physical differences are transparent, server differentiation becomes impossible and branding irrelevant.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

I am not a betting man (I paid attention in class when we covered probability) but I would bet that virtualization will change everything.

I concluded last week's column with the prediction that virtualization would make the server business more cutthroat than it already is.

My reasoning was that when hardware is isolated from the operating systems to the extent that it is effectively generic and any physical differences are transparent, server differentiation becomes impossible and branding irrelevant.

My second thought about virtualization is potentially more disruptive: A virtualized infrastructure weakens Microsoft's hegemony.

Note that I'm not taking Microsoft's foray into virtualization, Microsoft Virtual PC, seriously. When I talk about virtualization I'm talking about the likes of VMware. The biggest difference between Microsoft's product and VMware is the latter's support for non-Windows operating systems, notably the 57 varieties of Linux, as well as NetWare and exotica such as Oberon, Inferno and BeOS. Virtual PC, on the other hand, is really more of a migration tool for NT 4 shops.

In the Microsoft world, the intimate relationship with the underlying hardware makes the operating system special - Windows is "welded" to the platform, and the complexity of the resulting system means you have to be committed to the installation.

But virtualization changes everything. Now operating systems become as mobile and malleable as applications - in fact operating systems become, in practice, even more mobile than applications, and can be duplicated, relocated and reconfigured with amazing ease.

So if Microsoft doesn't own the platform any more, it puts the company in a weaker position. For example, where you need extra service and don't want to spend more money on Windows, you can run an alternative operating system simply configured for one or two services. This really will rattle Microsoft's ongoing spin-doctoring about the total cost of ownership of Windows compared with Linux.

If you take a Windows shop and try to migrate to Linux the cost will kill you, but that isn't reality. Reality is that you would use your existing investment and over a long period introduce Linux piecemeal. With virtualization, the migration process becomes far more manageable and therefore cost-effective.

I think the whole cost campaign Microsoft has been running shows just how profound a challenge Linux is to the company. What it is admitting is that there is a real pretender to the throne. Microsoft is, in effect, validating the competition.

This reminds me of many years ago when Novell did a "head to head" test against Banyan and Microsoft. The problem was that in most areas there was no clear winner and when you added it all up there was no overall winner. Not only was the effort a waste of time and effort, but Novell publicly admitted there was competition. Bad idea.

So with virtualization Microsoft loses an important advantage. What can it fight back with? Well, how about licensing? Could Microsoft start linking licenses to physical processors, for example, allowing only one server license to each real processor?