Me and my colleagues are thinking about how to solve a rather unique situation /at least in my experience/.
Our customers have been complaining about how fast [ how slow, to be more precise ] our application processes files [each about 1kB usually ] ... so we've spent some time inspecting the code of the would-be-thread and added a few lines for log4j to measure time between every operation to find out on which parts to focus. Up to this point it's all normal. The strange part comes that EVERY part to which we logged the starting and ending point got a LOT faster [ try 20 times faster ] normally it would take 4-10 seconds, depending on what else was running on the machine, but after adding the logs it suddenly dropped to 0.3-1.6 seconds per file.
Does anyone have a notion as to what we're looking for? We can't seem to find anything.