In my recent overview of Virtual Desktop Infrastructure technology, which included a review of pacesetter Citrix XenDesktop 2.1, I said IT managers shouldn't be rushed into making use of VDI. A reader commented on the XenDesktop 2.1 review with an even sharper "VDI--No Thanks" posting that adds further detail to some of the drawbacks I pointed out in my story package.
John Thompson says in part:
"VDI is a solution looking for a problem. It costs too much, even at the cheap price Microsoft sells it for in the Desktop Optimization package. [I didn't test this offering--CS] It is far more complex to set up and manage than you know. It can't be used on all desktops with all apps so you wind up having to manage a VDI infrastructure and a classic fat client desktop infrastructure--just how much time and effort in IT will you save now that you have two desktop infrastructures to manage, patch and secure. Disconnected users are still a problem--you can't virtualize all apps and packaging apps in a virtual package for a disconnected user is 4x the complexity and time it takes vs. using a best of breed fat client packaging tool. You have doubled your security footprint and attack vectors--if you think you haven't then you don't understand security and what the bad guys are up to in the VT Hacking world. Not all users will give up their desktops that easy. ROI will be extremely difficult to prove vs. server VT."
The reader is correct in pointing out the serious costs, infrastructure changes (one source for the story pointed out that the source's organization was only able to implement VDI because it already had the underlying network QOS (quality of service) features turned on--a costly prerequisite) and IT process changes that are mandatory to successfully get a VDI pilot project up and running.
However, as I pointed out in my start-of-the-year column on the virtualization field, I believe that virtualization will be an important tool that successful companies will use to help them absorb failed competitors. And it is this use case (a term VDI vendors are extremely fond of) where it makes the most sense to consider a leading- or bleeding-edge implementation of this type of product.