Tim Mangan, the founder of
TMurgent Technologies, has
published a paper about the perceived performance of Terminal Server systems. This paper discusses the difference between tuning a system for maximum computational performance and tuning a system for maximum perceived performance. It was inspired by this discussion (
part 2 ) on the THIN list regarding context switches, thread quanta, and application mode vs. background mode server settings.
Basically, his paper talks about how you can spend all the time in the world discussing the fine points of server specs and computational performance, but all that really matters is how the server performs in the real world. I agree 100% with the ideas expressed in this paper, and I definitely feel that this paper is worth reading. In fact, it nicely compliments my earlier Terminal Server Performance Tuning Guide.