What follows is in my view a typical scenario of one of the common curses of the performance tester. A new version of a test object finds its way into my hands. After some twiddling with it to get it running, I do a first quick test run. Somebody – development, architecture, project management, whoever – gets wind of it and is curious about the results. Defensively, I try a delaying tactic: “It’s the very first run. The test environment wasn’t completely built up. We have to do more tests. We must at first confirm the numbers before it makes any sense to discuss them.” My counterpart is getting more inquisitive. He insists to be aware of all of my objections and that he just wants to satisfy his curiosity. I finally give in and tell the numbers. Shortly thereafter, I regret it deeply. Like an avalanche, my little piece of information has gathered a gigantic mass in no time and like a boomerang it is coming full circle, heading straight back at me.
(read more…)