Wow, well HonorGod has given you the first 7 questions to get you started. If you have not figured it out, the type of hardware you are running WebSphere on (or any application server for that matter) is only a small part of in the decision of how many users the hardware can handle.
Application/System design and function play a big part. We have 6 dual CPU dual-core Xeon servers with 16GB RAM running WebSphere that we hope will support 1000-1500 users. Oh, the WebSphere servers only do about 5% of the work for our application, 10% is done by 4 single CPU quad core Xeon servers and the other 85% of the work is done by a pair of IBM zSeries mainframes.
Main Topics
Browse All Topics
by: HonorGodPosted on 2009-02-24 at 04:45:33ID: 23720802
Wow, this is a very difficult question for a number of reasons, chief among which are:
Q: Does the Application depend upon, or make use of a database?
Q: If so, what kind, and has any analysis been performed on tuning the
application and its use of the database?
Q: If so, is the database local, or remote to the Application Server machine?
Q: If the database is remote, how remote? (e.g., how many lan segments,
bridges, routers, firewalls, etc, are between the AppServer & the DB)?
Q: Have any performance measurements been performed of the current
environment (i.e., AppServer & Application)? Has any tuning of this
environment taken place?
Q: Is caching enabled anywhere? Would caching help reduce the need to
access the application? Where would caching benefit the overall
throughput?
Q: Have you investigated any performance or modeling tools that might
assist you to have a better understanding of how to plan your environment?
e.g., This paper is a little dated (12/2002 but it does contain some
information about how to approach this topic)
Title: High-Volume Web Site Performance Simulator for WebSphere
URL: http://www.ibm.com/develop
Hopefully this help to give you some ideas...