Why Not Desktop Virtualization?

Implications: Desktop virtualization promises huge savings in TCO, improved security, good user experience, improved manageability, etc.  Despite these advantages, it has many issues that have so far prevented widespread adoption, and likely will always limit adoption. 

Analysis: VMWare, Ctrix and others claim desktop virtualization is the answer to IT's prayers for desktop TCO and security.  Server virtualization has been taking the IT world by storm, so why hasn’t desktop virtualization done equally well?  Is it just because the products aren’t as developed yet?  Is it a public perception issue?  Or are there inherent issues in the model that limit adoption? 

The supporters of desktop virtualization talk about no longer needing to update desktop hardware, better compliance and data security, low TCO for desktops, and the ability to access the desktop image from any machine anywhere.  These sound like compelling advantages, particularly the low TCO and better security.  However, these advantages have been available for 30+ years.  It’s called a mainframe and terminals.  In the Eighties the users migrated en masse away from the mainframe/terminal model, as soon as PCs became available.  Larry Ellison got a lot of press in the late Nineties with his “thin client”, which is essentially the same model.  What has changed now to make the mainframe/terminal (oops, desktop virtualization!) model more attractive?

The short answer is:  Not much!  Users made it very clear in the Eighties that the overriding consideration was the quality of the user experience.  Most users want the independence and control provided by the individual PC.  They want the vast quantities of storage available, the quick response from the screen and keyboard, and the fast (relatively) processing power.  They want these things enough so that all IT’s arguments about TCO and security fall on deaf ears.  Desktop virtualization supporters claim to be able to match the user experience while providing all the benefits of a centralized solution.  Is this claim valid?

There’s not enough space here to fully investigate, but a few claims can be considered.  One big argument for desktop virtualization is to maintain one copy of the OS and push it so all local machines.  This requires a hypervisor running on all local machines that can interface with the OS copy.  That will likely be available at some point, but it’s not generally there today, and few companies are going to replace all their PCs to get it.  Said hypervisor and OS must be able to manage a plethora of hardware, from the very old to the very new.  If that task were easy Windows today would work lot better.  A significant percentage of the problems with Windows are tied to managing a wide variety of hardware.  Desktop virtualization won’t make that problem go away.  It just divides the problem between the hypervisor and OS. 

Backing up local disk storage is a thorn in the side of every IT department.  The total mass of storage is so great it’s virtually impossible to provide enough backup for all of it.  Moving it into a central virtual image doesn’t help – it just moves the storage from cheap desktop disks to expensive data center disks.  That increases cost, which means users have to be carefully limited on the amount they can use.  But that flies in the face of user independence.

Well managed environments today control their endpoints from a central location.  Patch management, updates, virus checking – all these can be centrally managed while still providing users the full experience they desire.  For most users, the incremental gains in TCO and security will not be worth the impact on the user experience.  There is a subset of users, typically those who use the PC for a limited function (e.g., clerks, word processors) for whom desktop virtualization will work and be a fine solution.  For the typical knowledge worker, however, it seems the independent PC/laptop is likely to live on.