Windows 7 hit nearly two years ago, but many enterprises have yet to migrate to the new platform. The organization
I work for happens to be one of them, and several months ago, my team began working on an enterprise desktop redesign.
We needed to simplify our desktop environment while providing greater security for our data and migrate to Windows 7. As part of the desktop redesign, we hoped to implement a virtual desktop infrastructure (VDI), but I soon came to the realization that our network could not handle it.
In conjunction with the desktop virtualization portion of the redesign project, we planned to implement user virtualization, application virtualization and application hosting. Unfortunately, we also ran into some time consuming issues when trying to virtualize tricky apps using popular application virtualization tools.
The application virtualization tools we evaluated include Microsoft App-V, VMware ThinApp, and Citrix XenApp.Microsoft App-V was a no-brainer for us because we had to pay very little to add the Microsoft Desktop Optimization Pack (MDOP) to our Enterprise Agreement. ThinApp had some interesting use cases, but we were not willing to spend the money on the product, nor did we want to complicate the environment with too many virtualization products. Our XenApp proof of concept turned out very well and proved to be another solution to get us closer to our desktop redesign goals.
Virtualizing Microsoft Office, Apple iTunes
We started out with a list of over 350 applications, and the first application we introduced via App-V was Microsoft Office 2010. There were several reasons we chose to virtualize it, the most critical being the need to keep Office 2003 on the local machines.
Microsoft had a great deal of documentation on how to virtualize Office 2010, but we ran into many hurdles along the way. Though we implemented it, I am very happy we did not make it our primary application for Microsoft Office work because we have experienced a number of issues since the deployment.
Since I planned an iPhone deployment, the next application we sequenced was Apple iTunes. Unfortunately, one of the engineers tasked with sequencing iTunes in App-V spent 16 hours working on sequencing the application and simply couldn’t get it to run properly.
The day we started sequencing iTunes, I happened to read an article by virtual desktop expert Brian Madden where he discusses a tool called App-DNA AppTitude that analyzes application compatibility with Windows 7, Microsoft App-V, Citrix XenApp and other products. It looked to be a perfect fit for what we hoped to accomplish.
I had estimated the amount of time it would take to evaluate and deploy applications in a Windows 7 environment, and the timeline was unacceptable. It took 16 hours of an engineer’s time to prove that iTunes would not work with App-V and by my calculations -- even after we cleaned up some of the redundant applications we had -- it would take about 36 months to complete the project. So we contacted App-DNA to get a product demonstration.
When we set up the first demo, App-DNA asked us if there were applications we wanted them to evaluate. We chose a handful of our internal applications, including iTunes since we knew how much time we’d spent trying to make it work.
A couple days later, they provided us with their reports. The information was very good; not only did it tell us where the software was a best fit, but it also told us the remediation steps necessary in order to get our software working in the different environments.
Within five minutes, AppTitude provided me with a report for iTunes showing us that we would need to remediate 268 different compatibility issues to get it to work with App-V. Sixteen hours of an engineer’s time would have been saved if I would have had the product before attempting the install. Now I know we need to deploy iTunes as a thick install.
The evaluation also included information on an application that we sequenced in App-V and begun to pilot without issue. When we received the reports back from App-DNA, it showed issues with this particular application when implemented with App-V.
It turned out that the users did not fully test the application or they would have ended up getting an error when they attempted a specific function in it. The AppTitude report also allowed us to remedy the issue quickly and gave us more confidence that the application would continue to run.
At that point, we made the decision to purchase the product. App-DNA has several licensing models for AppTitude, and we chose to license all modules in a three-year enterprise licensing agreement under a “business as usual” model.
I did this because we will utilize AppTitude as part of our vetting process while introducing new software into our environment. I also wanted to make sure we can introduce any software updates into the evaluation process to expose issues prior to implementing the update.
Also, to ensure that new software fits into our architecture, we can use this tool to vet the application up front before we purchase it. As part of our project approval process, we get the executable for the application and run it through AppTitude before we approve the project.
We have in-house developers who also take advantage of AppTitude as they develop applications. This cuts time to market for applications, as AppTitude provides early information regarding software compatibility issues. At this time, we run each executable against all of the modules that apply to the particular application, and this includes: IE8 compatibility, Windows7, App-V, Citrix XenApp streaming and hosted, Server 2008/ 2003, and interoperability.
Since introducing AppTitude about two months ago, we have imported over 90 applications and successfully implemented about 20% of the remediated applications into our environment. One of our biggest struggles has been actually gathering the executable for the applications being deployed.
We now recognize more than ever that we did not have a good handle on our application environment. Some applications have specialized .ini files that track different things, such as workstation ID and printer settings. These one-off anomalies can show false positives in the software. The key is making sure that there is in-house knowledge about that application so these anomalies can be understood.
Of course, the App-DNA product can only go so far. In a couple of cases, we got the green light to deploy applications but when we implemented them, we received errors. Up to this point, we have been able to resolve all of the issues, but it demonstrates that the product is not a magic bullet. It is as close to one as I have seen though.
ABOUT THE AUTHOR:
Jeff Moore is an IT Manager in financial services with nine years of experience in research, development and infrastructure planning.