Could the stretched-out replacement cycles for desktop machines be a boon for client computing? In a recent Wall Street Journal Business Technology blog post, Ben Worthen noted that a survey from our print publication, CIO Magazine, found that companies will forgo traditional three-year replacement cycles for desktop machines (both traditional desktops and notebook computers). According to the CIO survey, 46 percent of businesses will defer replacing machines for the next year or two.

Worthen says that this will be a problem for people who are already suffering from overloaded machines, bogged down by big apps and too much data. I'm not so sure about that. Any machine purchased in the past three years should be capable of holding at least 2 gig of memory, which should be plenty for most people's workloads. On the data side, most three-year old machines should have at least 40 gig of storage, and probably more. It's hard to imagine most work environments requiring more than 40 gigs of storage.

However, I think he's on to something—not so much from today's workloads, but from tomorrow's. Specifically, the looming (semi) forced shift to Vista or Windows 7. Both of these versions of the OS require a significantly larger hardware footprint than XP does. Consequently, there's a collision course between the operating system of the future and the hardware of the present—which presents an enormous opportunity for client virtualization.

There's three ways that client virtualization can help out in a capital-constrained environment:

If the current hardware really is overloaded by heavyweight apps, presentation virtualization is a possibility. This technology puts the application back on the server and merely shunts the user interface out to the client machine. Instead of having to host the entire application process and store the data, the machine acts as a rich client.

If the client hardware is insufficient to run Vista or Windows 7, move to a Virtual Desktop Infrastructure (VDI) environment, with a virtualization server hosting multiple desktops. There's no need to outfit end point machines with 4 gig of memory and 200 gig of storage. It's not even necessary to scale that level of resource onto the server.

For example, if you have 10 desktops on a single server, you don't need 40 gig of memory. Because end user machines are very spiky in terms of usage and, frankly, under-utilized 99 percent of the time, a smaller amount of resource is required on the server. In other words, the resources can be multiplexed. This is financially savvy for two reasons: (1) Due to the multiplexing effect, you don't need to buy as much total resource capacity as you would if you were provisioning individual end points, each with sufficient capacity to support a Vista environment; and (2) buying in bulk for servers is, up to a certain point, less expensive than buying the same amount of capacity for individual end devices—in essence, you're paying wholesale rather than retail (so to speak) for hardware capacity. There is a finite ability to play this "wholesale vs. retail" tradeoff; typically when you start putting very large memory sticks into servers the price for it escalates wildly.