> My experience with commercial load-testing apps is that they are > outrageously expensive, a pain to program, don't really scale all that > well, and mostly have to run on Windows with someone sitting at the > mouse. There are some that work better than others, but the free stuff > in this areas is quite good.
Ditto. I've found it easier to hack together something with LWP's LWP::UserAgent and Benchmark. Particularly when load-testing an application that required different pre-registered users that had not performed this particular transaction we were load testing. (A survey with random and different questions - depending on the user!). None of the macro-like testing apps could do very much with regard to that kind of interaction and variability in the content generated by the application we were load-testing. But a 30-line Perl script that simply appended the Benchmark results into a tab-delimited file worked great. We found about 30 instances of Perl running the script per WinPC ate the machine. So we only loaded 15 per PC during actual testing and added more distributed nodes to the test. As an aside, the whole thing was an exercise in needing a cup of sugar and asking the local grocery store how much sugar they have on the shelves. i.e. What is the point of measuring beyond the "more than the 1 cup you need"? So we measured (at great expense) and determined that the entire lifetime load (~1yr) expected for all users on their system could be accomplished on the existing sytem during a lunch hour. Mike808/ --------------------------------------------- http://www.valuenet.net
