Desktop virtualization \u2014which I dissed in a previous blog entry as being applicable in only relatively limited circumstances\u2014is showing a lot more resilience and adaptability that I gave it credit for. Desktop virtualization vendors have come out of the woodwork to talk about how their particular product sets \u2014many of which have not traditionally been described as virtualization of any kind\u2014are actually contributors to desktop virtualization. And the problem is, in many cases, they're right. "Desktop virtualization" just means separating the machine on which the user works from the software s\/he uses. That can mean running a whole session on a blade PC in a data center, running dozens of PC OSes as virtual machines on a server, running dozens of application sessions on a server and piping control for each out to a desktop, or any of half a dozen other interesting permutations. HPcalled to get me up to speed on how their blades could contribute to desktop virtualization projects and confused the hell out of me before I realized they weren't saying blade-based PCs were better than desktop infrastructures from VMware or HP. They were saying blade PCs were part of the desktop virtualization world that stereotypically involves PC operating systems running as VMs on a back-end server. A lot of customers don't want to run hundreds or thousands of PCs as VMs, but do want the cost savings and simple administration that companies can get by connecting each user to a dedicated back-end blade-based PC rather than a tower under their desks. HP's point\u2014at least the one articulated by its blade-PC people\u2014was that there's lots of room in the desktop virtualization market and they're in a good position to take at least part of it. Newcomer MokaFive'sapproach is to create virtual PCs in software, then download them from a central server to an end-user's PC \u2014complete with operating system, applications and whatever group policies the company wants to enforce. The LivePC then runs on a user's machine independent of whatever else is already running there, supplying a company-approved PC configuration that can be launched and used whether the user is connected to a network or not. MokaFive can also put a virtual PC on a flash drive and let you carry it around with you. Xenocode does a similar thing with just the applications \u2014ginning up a platform for Java and .Net applications to be installed and run on flash drives or PCs, without requiring much, if anything, of the hardware on which they run. Pretty neat, both of them; not remotely what I think about when someone says "desktop virtualization," but it certainly qualifies. The uncharitable interpretation of what's going on in the market is that the term "desktop virtualization" has ceased to mean much because it can me so many different things. The charitable \u2014and probably more accurate \u2014interpretation is that the desktop virtualization market is fragmenting quickly into a host of niche markets, each of which is "desktop virtualization" in the same way that mainframe system-software development and AJAX casual game programming are both "application development." They use a lot of the same tools and the apparent process (lock door, shove pizza underneath ). But what happens behind the closed door and the purpose to which the product is put are very, very different. Which is a good thing for end users, who should be able to pick from a menu of options to find exactly the set of functions that meet their needs and preferences. (It's also a good thing for bloggers and analysts, who are free to boldly declare desktop virtualization The Next Big Thing, the current Big Thing, the last Big Thing, the Thing that was never Big, the Thing that will never be Big, the next-generation Thing, or the Dead Thing. At any given time during the next five years SOME part of the desktop virtualization "market" is guaranteed to fit whichever resolution the Magic 8-Ball predicts.)My personal Magic 8-Ball predicts exactly the same thing. Desktop Virtualization is here to stay; I'm just not sure what kinds of it we're going to end up with in the end.