Can anyone explain the science behind the sampling functionality of visualvm ?
I would have thought that collecting the amount of CPU spent on each particular method would require instrumenting the application to be profiled, but it doesnt look like visualvm is doing any kind of instrumenting, so I'm curious to know how it's done...
It looks like the new VisualVM-Sampler plugin gather performance and memory data by periodically polling the monitored application for thread dumps or memory histograms. Check this article for more details.
When I start memory profiling JVisualVM spends a couple of seconds with instrumenting all the classes; at least that’s what it says it’s doing.
That sure looks like instrumentation to me. :)