I’ve only heard of CPU Flame Graphs since the article on NodeJS performance issues at Netflix.
… given a performance problem, observability is of the utmost importance. Flame graphs gave us tremendous insight into where our app was spending most of its time on CPU. I can’t imagine how we would have solved this problem without being able to sample Node.js stacks and visualize them with flame graphs.Yunong Xiao @ Netflix
The source code is available on Github.
It uses DTrace (or perf_events, SystemTap, or ktap) to capture the kernel output and convert it into an SVG image for you to analyse (see an example here).
Now all I need is an excuse (and time) to use this!