Analyzing workloads, cloud or otherwise, is something all engineers end up doing at some point or the other (or it’s literally their job description!). An obvious reason to do so is during performance analysis: it’s noticed that, say, the workload’s CPU usage spikes at times, causing issues, even outages.
The need of the hour: observe, analyze and tease out the root cause of said performance issue! Of course, that’s often easier said than done; this kind of work can bog down even experienced professionals...
On modern Linux, perf comes to the rescue! But (there's always a 'but' right), the report(s) perf generates can go into dozens of pages. We need a visual tool, the FlameGraph (FG)!
Will also demo a wrapper script I use to generate FGs; we'll also interpret perf's output via the super speedscope web app!