How can I evaluate and compare the effectiveness of remote desktop protocols based on the quality of their user experience?

Question

There are many remote desktop protocols used widely, e.g. VNC, RDP, PCoIP, RGS, etc. Taking a look at their specification, it seems that they provide different features, such as redirecting I/O, tuning display setting, etc. I've recently installed a thin client system at my company. By using RDP protocol, the users are complaining about slow update of their screens. PCoIP has also slowed down their IP telephony activities.

Now the Q is: How can I evaluate and compare the effectiveness of remote desktop protocols based on the quality of their user experience?

Answer 1

Here are my findings:

• RAWC by VMware
• LoginVSI by Login Consultants
• and DeskBench

Answer 2

I'd try to steer away from peoples "gut feel" for performance and get numbers to do the talking.

I use a variety of benchmarking tools to evaluate the performance. Since it's about user experience you need to consider the type of user, Task, Knowledge, Power user etc.

I can't address what is the "best" protocol as it's really a moving target and has many factors, but our baselines are in a XenDesktop environment with ESX etc. We have images with SAS storage and EFDs for cache drives.

Now the Q is: How can I evaluate and compare the effectiveness of remote desktop protocols based on the quality of their user experience?

Tools I use are..

OSMark

It's a great tool I customise the tests to suit the environment eg. CPU intensive, Graphic and so on. I can then compare rendering of RichText, Web and 3D objects when I make changes to the environment and relate back to the baseline. Youcan also run the same tests on a physical machine to look at the variance.

Crystal Mark

Benchmark disk performance, good for "Internal to VDI" testing of a VDI disk\network performance.

If you have a Citrix Environment, EdgeSight for Load Testing is great as well.

While this does not address your problem in particular it might help determine your bottlenecks and create load on the system for other testing.

You may need to look into QOS for your telephony stream and seperate the traffic once your determine what is most important. Look at if you can change compression in your telephony system as well.

Hope this helps, or is useful to anyone passing through.

jezr

Answer 3

I am looking for some tips also, but came pretty much empty handed.

The best stuff I am using to benchmark the performance of RDP-like solutions is WireShark and a synthetic test.

Run WireShark, connect and do some (ideally scripted, for reproducibility and comparability) operations your users would do (go to menu, edit setting, ave a fullscreen refresh of some picture that is hard to compress and then some window maximize and minimize to see how big monotone screens are compressed and so on).

Measure the time between the click that starts the refresh and the complete refresh showing up ( when the data stops flowing)

Watch out for things like clocks, widgets showing real-time data and other stuff generating a lot of refreshes every second, that can produce some noise making it hard to evaluate the results.