Would VDI please just go away?

Earlier this month, VMware rolled out yet another half-baked strategy for salvaging its much-maligned Virtual Desktop Infrastructure (VDI) initiative.

By creating a new remote access client (so that far-flung users can access their centrally hosted desktops), and then handing the code base over to the open source community, EMC’s VMware unit hopes to jump-start interest in so-called “appliance computing” among the most rabid segment of geekdom: the FOSS (free open source software) crowd.

Personally, I’ve never been a fan of VDI. The idea of consolidating desktops to lower TCO always seemed problematic to me for a number of reasons:

1. End users hate it. Users don’t want a centralized, server-hosted workspace with lots of quirks and limitations. What they want is more flexibility and mobility in accessing their applications and data. The former is taken care of by extending traditional, “fat client” OS and applications into the cloud, while the latter is why the latest generation of low-cost notebooks and netbooks is so popular.
2. It doesn’t scale. Hosting large numbers of virtual machines is a demanding proposition, requiring lots of big servers with gobs of memory and storage. It’s expensive and not particularly cost-effective from a hardware investment standpoint. As a colleague of mine at one of the world’s largest financial services firm puts it, “VDI is a great way to turn a 250-user Citrix server into a 25-user VMware server.”
3. It’s overkill. Many of the problems VDI purports to solve are better addressed by less radical alternatives, like application virtualization. In fact, there are very few scenarios where VDI shows a clear-cut advantage over traditional compute models. Information kiosks come to mind, but even these can be better served by a more scalable multi-user platform, like Citrix XenServer.

Basically, VDI remains a solution looking for a problem — a way for VMware to expand beyond the datacenter by squeezing every conceivable desktop computing scenario into a virtual machine-shaped box.

The company would have been better served by more tightly integrating its already robust VM architecture with Windows Vista and Windows 7. There’s a genuine need for a seamless legacy compatibility layer in these versions, and so far Microsoft has dropped the ball with its anemic MED-V offering. VMware could have replicated its successful Fusion model from the Mac, leveraging ACE, ThinApp, and other neglected technologies to create a solution that would have endeared the company to IT while embarrassing its arch-rival.

So, I say enough already with the worn-out delusions of world domination through virtualization. It’s just not going to happen, regardless of what those ivory tower types say over at Gartner or Forrester. After all, these are the same folks who predicated the utter failure of Windows NT 5.0 (a.k.a. Windows 2000) and who envisioned a world full of Windows CE-based terminal devices all connected to Microsoft’s Terminal Services.

Now, excuse me while I fire up my quad-core, 8GB Dell Precision M6400 laptop running the third-generation descendant of that same traditional “fat client” OS the analyst crowd wrote off nearly 10 years ago.