Services keep getting smaller and more distributed, deployed across many container instances in different nodes and regions — which makes profiling, analysing and improving performance considerably harder.
Continuous profiling swaps point-in-time data collection for continuous collection and analysis, aiming at a near real-time understanding of application performance across the whole stack. This is a pragmatic look at how it works and where it fits.
