This document presents guidelines and a general approach for evaluating the capacity of a system in the context of a specific deployment. Most of the key recommendations are also illustrated with examples based on a few scenarios that use Microsoft® Office applications. The document also provides guidance on the hardware and software parameters that can have a significant impact on the number of sessions a server can support effectively.

This whitepaper is interesting for several reasons. One of them is that it is the first hardcore 64-bit only whitepaper (since R2 is 64 bit only). It presents some interesting results, like this:

On of the things that is good to see is that MS also tested the effect of hosting RDSH session with all the Aero bells and whistles enabled. You know, the stuff that you never want to enable because it kills your scalability right? Wrong! Since this is all about (aero) remoting, the performance hit is not that bad at all. Not on the server side at least. You do of course have to keep bandwidth (LAN/WAN) and local client CPU in mind.

Throughout the whitepaper Microsoft mentions about 4540 times that these numbers are "experimental" primarely because everything depends on the application payload executed in the sessions. They are very right to do so. This just tests Office application while you company will most likely have at least one "special" LoB application that does not fit this bill. Still, it is a very nice guideline to start with.

Get the whitepaper here: http://www.microsoft.com/downloads/details.aspx?displaylang=en&FamilyID=ca837962-4128-4680-b1c0-ad0985939063