Bring-Your-Own Laptop Won't Help VDI

When I made my 2011 predictions, I didn't hold out much hope for virtual desktop infrastructure (VDI). I don't think that the upsides in moving to VDI--fast ROI, reduction in operations expenses, security enhancements, etc.--are enough for IT shops to change what they are doing. And I mean any type of VDI, from the traditional remote desktop that Citrix and Microsoft have offered for years to the newer VDI products that run as virtual machines on a desktop or off removable media, some of which we reviewed in 2009. One of the drivers, as Forrester Analyst Andre Kindness notes on Twitter, is that bring-your-own computer is forcing companies to VDI. I disagree, not because you can't use VDI in that situation--if you are going to support bring-your-own laptop, then VDI would be preferred for company apps--but because the costs associated with supporting bring-your-own laptop and transitioning to VDI for most companies are too high compared with the organizational benefit that would be gained.

The idea of bring-your-own computer isn't new. In the last eight years or so, laptops have gotten more powerful and prices have dropped in the dirt. Most people can afford them, and as far back as 2005, Current Analysis research showed that laptops sales surpassed desktop sales. It's natural to think that in a fashion similar to how IT was forced to support mobile devices like Palm Treos, BlackBerrys and iPhones because users were bringing them into the office, IT will also have to support employees' personal laptops. In a lot of companies I talked to, mobile devices that connected to Microsoft Exchange and carried documents were grudgingly allowed, often with no formal support from IT, and it didn't really cost IT anything tangible to let users connect to Exchange. If your company wants you to have a smart phone, in most cases they buy it, provision it, and give it to you. Some companies may actually allow employees to use their own laptops, but I don't see it being a trend. More like a tolerated anomaly.

There are couple of fundamental issues with bring-your-own laptops that are unlike the oft-compared mobile devices. The first is that any decent IT shop has spent a long time putting in place an environment that ensures a stable and reliable mobile computing platform. That's the goal, anyway. A bring-your-own strategy takes an IT process and tosses it out the window because desktop virtualization, running a hypervisor on your own laptop, adds its own complications to desktop stability and usability.

The host OS can adversely affect the virtual OS in ways that can be hard to pinpoint and difficult to resolve. I have been doing desktop virtualization using VMware's Workstation and VirtualBox daily for more than a year, and at times I have had keyboard and mouse input disappear. I have had the desktop video go blank. I have had installed USB devices disappear. Often times, I wasn't able to pinpoint the cause of the problem, and the resolution was power cycle off the VM and the host. There was one case where a network driver didn't play well with VMware and caused Windows 7, my host OS, to blue-screen. Not fun considering the lost work and downtime.

That leads me to the second fundamental issue. Any IT admin who has been on the front lines of support knows that once you drop software onto a user's personal computer, IT becomes responsible for all support, regardless of whether the software IT deployed caused the problem the user is having. Unfairly or not, if you put software on someone's computer and something breaks, IT has to fix it or explain to someone why it isn't IT's problem. (As if the latter will go over well.)

It's fine to say in theory that a desktop hypervisor running a VM makes a clean separation between the two computers, but, in reality, they are tightly intertwined.For example, using VMware Workstation, if I plug in a USB device and it attaches to the VM, it doesn't appear in the host laptop. There may be a way to fix that. I haven't looked, but it's pretty weird. If the host laptop is starved for memory and starts swapping to disk, the whole computer is going to crawl to a stop. If some application in the laptop or the VM decides it wants to do something like indexing the drive or running an antivirus scan, that is going to adversely affect the performance of both the physical and the virtual computers. I could go on for pages, but you get the point. IT has to be prepared to deal with a new set of problems cropping up.

Granted, all of the IT management issues, and others, that I raise are surmountable with a change in how IT manages desktops and a resetting of the organization's employees' expectations of desktop behavior and support. The newer crop of VDI software products tries to address the separation of host from VM, provide central management, support connected and disconnected use, ease updates and ease other IT management burdens. However--and this is my third and final point--I question if those tools are enough to allay the very real concerns that IT administrators have in deploying and managing a VDI environment on user-owned hardware.

For most existing organizations, changing from an environment where IT procures, provisions and manages laptops to an environment where they drop virtual desktops onto users' laptops is a radical change that is full of potential pitfalls and unknowns. What's the upside? Better management? With modern desktop management tools that offer full desktop control, remote control, patch management and a host of other features, desktop management hasn't been simpler. IT departments are going to have to manage desktops anyway, so why not include laptops in the mix, as well? In many cases, that is going to be far more cost effective for existing organizations than rolling out a VDI plus a management system.

And I haven't even bothered to talk about the security and governance implications that are a whole 'nother set of issues. If companies are going to adopt VDI, it's going to take some very, very compelling reasons for companies to change what they are doing, learn how to support these products and resolve end user desktop problems. (Try telling an employee that he or she has to back-rev a driver because it conflicts with the VDI software.) Then there are the hardware and license costs for the VDI software. That is a pretty tough sell. It's easier for organizations starting out and building an IT shop to adopt VDI from the start, and that is why I think VDI adoption will continue to be slow.