Improving screen redraw rate on thin-client over RDP to Server 2008 R2

Question

I've been reading about generally increasing the performance of an RDP session here but those posts don't seem to cover what I'm looking for which is specifically the redraw rate of windows. I'm not interested in video playback or CAD/graphical performance.

I have Wyse V10L and C10L clients connected to a Server 2008 R2 host. It's a virtual machine and it has enough horsepower to handle the client load. I don't believe the poor screen redraw rate has anything to with the host itself.

From my Win 7 machines using Remote Desktop Connection I get a fairly decent redraw rate that does not leave me feeling uncomfortable or annoyed. However, on the thin-clients the redraw rate is much lower and there's a big difference.

I've tried disabling compression, USB/serial/sound, and pretty graphics (font smoothing, backgrounds, window dragging, etc.). None of this has made even a small difference in the redraw rate.

Is it as simple as the Wyse hardware not being able to redraw any faster? I don't have a more powerful Wyse device to compare with. Any ideas?

Answer 1

Performance issues like you're describing are almost always (in my experience) network related. RDP sessions (in my opinion) are much more sensitive to network issues such as lateny, congestion, and packet loss than they are to almost any other resource issue (memory, CPU, etc).

That being said I've seen enough cases with dumb terminal performance to know that dumb terminals often don't have the horse power to provide a well performing session.

Your own test seems to bear that out; a full fledged computer running Windows 7 doesn't seem to be affected by the problem so that leads me to believe that the problem is with the dumb terminals themselves. If you have a newer, beefier terminal at your disposal I would drop that on the network and see how it performs. I would also try another workstation or laptop to see if you get the same good performance as the Windows 7 workstation. If a more powerful terminal or another workstation exhibit good performance than I think you can definitively point your finger at the terminals.

As a final note, there's only so much tweaking you can do on the server and/or the network before it becomes a "war of attrition" and leads to an excercise in futility. If the terminals are the problem then the resolution is to replace them. I went through this recently with a customer and the final resolution, after all the poking, prodding, and tweaking, was to replace the terminals.