Benchmarking your software is often much more complicated than it sounds. Sure, it’s easy to get some numbers that say something about the performance of your software, but if it wasn’t calculated using a very accurate representation of the actual patterns of use by your end users, it may be completely different from the actual results you get in the wild. . Websites are generally difficult to compare. Of course, you can run a script that measures the time it takes to create the page, but it will be a completely different number from what you see when used in the real world.
To create a solid benchmark for what your servers can handle, you first need to figure out what your users' usage patterns are. If your site is already running, you can easily collect this data from your logs. Then you need to create a simulation that will emulate the same templates as your real users ... this is to view the start page, log in, view the status page, etc. Different pages will create a different load on the servers, requiring the actual selection of the correct set of pages when simulating loading on your servers. Finally, you need to find out what resources are cached by your users, you can do this again by looking at your access log or using a tool like firebug.
source share