App-V Performance Best Practices: New Project VRC White Paper
Project Virtual Reality Check (VRC) released a new white paper titled “Impact of Microsoft Application Virtualization (App-V) 5.0, Optimizations and Best Practices” last week. This paper contains detailed Microsoft App-V performance data which can be applied to any App-V infrastructure scenario. As a member of Project VRC who was deeply involved in the many Login VSI tests that went into the paper, I would like to share some highlights with you in this blog.
If you are not familiar with Project VRC, it is an independent R&D project focused on the desktop and application virtualization market. The group has published several white papers on the performance and best practices of different hypervisors, application virtualization solutions, Microsoft Office versions, Windows Operating Systems in server hosted desktop solutions and the impact of Antivirus software. For the newest Project VRC white paper on App-V performance, my team member Microsoft MVP Ment van der Plas and I together performed many of the actual tests and offer our analysis.
Using Microsoft App-V or any other application virtualization solution will affect the performance of any system. This is because of the additional virtualization layer between the operating system and the applications. The best performing scenario (in a virtualized desktop environment) compared to native installed applications is when App-V packages are locally cached.
The graph above shows that in this test scenario, App-V 4.6 has a VSImax capacity impact of 8% compared to native installed applications. Meanwhile, App-V 5.0 has a 13% VSImax capacity impact. This difference of 5% is caused by the fact that App-V 5.0 has more and tighter integration options with the local operating system.
We also discovered during our research that Microsoft App-V 5.0 has a dependency with Microsoft .NET 4.0. The .NET Framework has a process that can have a seriously detrimental effect on performance, especially in a stateless VDI environment (see graph above). It is very important to always optimize the .NET Framework to avoid performance problems. Based on some of our earlier test results, Microsoft added optimizations into their best practices guide for App-V 5.0.
When using streaming, the SMB protocol performs best. Streaming applications has an impact of 6% (through SMB) and 15% (through HTTP) compared to locally cached applications. This offset is also noticeable in publishing times. We measured a small impact (4%-7%) on capacity when applying Registry Staging--both in cached as well as in a streaming scenarios.
The VSImax test results in this blog posts are only a small part of all the detailed App-V performance results and best practices in the white paper. This white paper can help any system administrator or IT manager to make the right App-V decisions. You can download the entire white paper for free at www.projectvrc.com/white-papers.