I’ve been investigating some of the OpenSource performance test tools available on the market, and it seems that most of them are lacking in features that are essential for performance testing in enterprise environments. A good list of both functional and performance tools can be found at www.opensourcetesting.org.
The biggest problem is that the tools are not designed to give out analyzable data in exported form, or even in viewable and understandable form on the screen. An example is soapUI that allows you to run functional test cases directly in multiple threads (to create load) but there is no advanced view of the results, nor can you export the results to anything for later analysis, thus making the feature mostly useless.
What I would need when I do performance testing on a system is a detailed transaction log of every request made from the load-generators point of view, and monitoring results from the target servers (CPU and memory minimum). All these results should then be combined together based on a global time-index to be able to compare them. Without this data it is usually very difficult to determine where the problems are in large-scale enterprise system where you have multiple web and application servers + database servers.