Real Costs of App Virtualization
There are three parts to determining if the money saved by using application virtualization is less than conducting business as usual using conventional application management tools. First, determine how much IT time is used today to conduct regression tests to ensure that new or updated software will not interfere with already installed software, which could impinge on ROI. IT managers next need to determine the amount of support time-along with lost user productivity-that is needed to remedy the inevitable problems that are missed in regression testing. More difficult to measure, but worth considering as part of the equation, is lost productivity attributable to delayed implementation of new products while this testing takes place.The dirty secret, if you will, of application virtualization is that there are other costs that must be factored in to the overall cost of the application virtualization project. Applications must still be packaged or sequenced and prepared for virtual deployment, just as in the days of old. Applications must still be tested to see if they work correctly when virtually affixed to the large variety of end-user systems where they may land. And, although application virtualization vendors are loathe to admit it, there is some sort of software agent that must be present on the end-user system for all this technology to work, so figure in the (often minimal) cost of deploying and maintaining the agent. While the agent is often not installed (though sometimes it is, as is the case of Microsoft's App-V), there is an agent. In many cases, virtual application products will use existing software deployment tools to provide virtualized applications-or components of virtualized applications, such as icons or desktop shortcuts. While this is rarely a requirement for using application virtualization, the per-seat costs of the software deployment solution must be factored into the cost of the application virtualization tool. Also, keep in mind that providing centrally managed applications on demand requires plenty of reliable network bandwidth. In addition, when using the infrastructure platforms, such as Microsoft System Center, Novell's ZENworks or Citrix's application delivery infrastructure, there is usually plenty of hardware, licensing and implementation planning needed to support secure, reliable application delivery. My testing experience has shown that the rather costly and cumbersome management of application virtualization assets should be done much earlier in the implementation process than with server virtualization. The isolation techniques used by application virtualization can make it easier than relying on Windows to keep new applications or updated versions of already deployed applications from interfering with each other. In many cases, some applications need to work together. Overcoming this isolation so that applications can interact with each other is where the biggest recent advances have been in application virtualization. A challenge that lingers over application-and even desktop-virtualization that is significantly different from server or storage virtualization is that of disconnected operation. Applications need to be used everywhere, including off the network. Application virtualization vendors are working to overcome the nontrivial challenges of taking applications off the network while still providing a good method of synchronizing changes when reconnected. Right now, some platforms, such as Citrix and Novell's ZENworks, enable users to use applications offline. There is a significant amount of additional work and infrastructure needed to make this happen, and it represents a significant technical challenge for IT administrators. Offline operation also raises significant challenges for the enforcement of security and data use policies, which is likely a business driver for pursuing virtual application delivery. Technical Director Cameron Sturdevant can be reached at firstname.lastname@example.org.
Finally, determine how much support time and productivity is lost when users muck up applications through misconfiguration, and add the amount of time and labor required to recover end-user systems because of viruses and other security compromises.