Perhaps it is a truism of IT that technologies take far longer to be adopted than we would ever have expected, but that has certainly been the case with application virtualization.
Softricity, now part of Microsoft, brought the technology to general attention in the early years of the decade. They were joined by other innovators such as Altiris (now part of Symantec) and ThinStall (now part of VMware), who jointly moved the technology forward. Expectations were high that adoption would be swift.
Looking back it is now clear why adoption has been slower than expected, and why we could now be at the point where broad adoption really takes off.
But first some background. In Microsoft Windows, we have gotten used to applications arriving in .msi software packages that install automatically. The package knows how the application should be installed, queries the machine to understand what needs to be written where, runs through all the necessary changes and, where necessary, updates system files that configure the application and the system.
If that machine had only that one application, then the situation would be simple -- we would have a system with exactly the right configuration for that application. The reality is that there are many applications in a business, many but not all of which are present on any one machine. This makes it impossible to know if the configuration and system changes that one application makes are going to affect other applications. This is the fundamental problem that has dogged our ability to test systems effectively.
Application virtualization sets out to fix this problem. When an application is delivered through application virtualization, it does not get installed in the way we would normally understand. The application is made available to the user and the operating system, but it's isolated so that its presence does not change the underlying system files. By doing this, applications can be used that otherwise might have changed the operating system in undesired ways. This eliminates the problem of application conflicts and makes it possible to have an application appear and be used and then, potentially, disappear leaving no damaging footprints on the system.
The issue that dogged application virtualization in the early days was that of application compatibility -- sometimes applications either would not work in an isolated 'bubble' or other applications would not be able to work with the virtualized application. With such a fundamental change in the way applications work, it took time and practical experience to understand exactly what needs to be isolated and what left open and hence move application virtualization products forward.
Today most applications can be delivered as virtualized apps. Vendors continue to work to increase the proportion of applications that can be virtualized, but we are beyond the point where application compatibility would be a roadblock. So much so that many of the organizations I talk to are now making their decisions on which product to choose and whether they should be delivering all the applications in the business in this way. But in some ways this is a distraction from more fundamental change that may be happening: software vendors distribute their applications already packaged for application virtualization. Microsoft has made the first move in making 'Office 2010' available already packaged for its 'AppV' application virtualization product. The prospect is that other vendors will follow suit and make their applications available in this form too.
As you look at application virtualization you will need to consider how to manage the user environment. The point here is that the isolation bubble also isolates the user environment and, unless managed appropriately, the helpdesk will find that settings and data associated with the application are not where they would expect, making applications difficult to support. However the change we are looking at here would be the most fundamental shift in the way that applications are packaged since the arrival of installer technology in Office 2000, way back in 1999. This change has the promise of solving the problem of testing of client systems.
Martin Ingram is the VP of strategy at AppSense, a leading provider of user virtualization technology to enterprise organizations.