With Thanksgiving and Christmas behind us and the New Year upon us, it is time to take stock and see what changed for desktop virtualization in 2010.
One thing is very clear: We have moved from desktop virtualization being 'about to take off' to 'has taken off' -- the evidence for this is pretty clear in the number of licenses sold. With volumes sold in the low millions, desktop virtualization is way beyond the tryout and pilot stage. However, desktop virtualization is not yet for every user. There are a number of areas where it's still not a good fit.
For example, a user who does not always have access to the Internet from his or her laptop may not be able to use a hosted virtual desktop. This is a problem that will be addressed by client hypervisors in coming years. These provide the management benefits of desktop virtualization to the intermittently connected user. For now, they are very new but will become critical for mobile workers and may also have a major role to play in bringing down the costs of desktop virtualization for non-mobile users as well.
Of perhaps more concern is the question of the economic basis for hosted virtual desktops. This has recently become more visible thanks to Microsoft's paper 'VDI TCO Analysis for Office Worker Environments,' which compares the total cost of ownership of traditional PCs and their virtual desktop alternatives. Their conclusion is that hosted virtual desktops are more expensive to deliver than a traditional, well-managed PC. There are a number of interesting points and conclusions to draw from this document.
Firstly, Microsoft compares virtual desktop infrastructure (VDI) against a 'well managed PC.' This is a key phrase for those of us who work with industry analysts. It means far more than just having policies and necessary technologies in place: It assumes that the PC is locked down to the extent that users can make no changes to the configuration of the machine, including being able to install applications. While this is the most cost effective and secure way to manage a PC, it is not a good solution for many users; either they culturally expect to be able to adapt 'their' PC as they see fit or they actually need to be able to install applications beyond those provided by IT.
Secondly, the target user base is an 'Office Worker' - the most general and easily standardized user. As with any new technology, it will not be right for everyone from the get go. General office worker's are not the most appropriate target for desktop virtualization at the moment, partly for the economic reasons that Microsoft sets out. In the case of desktop virtualization the ideal users today are those who are expensive to manage conventionally. These may small groups of users in remote locations or those with pressing security requirements. In the case of the remote user groups, significant savings can come from the reduction in desk-side support that can be made by centralizing execution. Security benefits are always hard to quantify, but being able to keep all data within the data center by centralizing execution and delivering only a display feed to the user can be critical for some government, finance and healthcare projects.
Thirdly, while the report shows that hosted desktop virtualization is more expensive than a traditional PC, the reason that it is more expensive is due to higher software licensing costs in desktop virtualization -- principally Microsoft software licenses. In other words, they are both showing that desktop virtualization is more expensive and that they have control of this price difference. When and if Microsoft decides that it is in their interest to change the economics of hosted virtual desktops, then they can flip the switch and change the game.
Whichever way round this plays out looks good for desktop virtualization: Either hosted becomes cheaper or the swing to client hypervisors is faster. This is gratifying to see in light of all the work currently going into early implementations -- the work will not be wasted.
Martin Ingram is an independent commentator on the virtualization space.