The fourth pillar of observability: What is profiling and how does Grafana Pyroscope work?

Observability is based on three pillars, the so-called “metrics, logs and traces”. Metrics tell us that something is wrong (e.g. a spike in CPU usage), logs help us understand what happened at a specific moment, while tracing shows which service in a distributed system is the bottleneck. Recently, however, a fourth level of detail has been added to these three pillars - profiling. It tells us exactly which function, and even which line of code, is consuming specific resources.