Beyond monitoring
Observability is the ability to understand a system's internal state from its outputs, so you can answer questions you did not anticipate. It rests on three complementary kinds of data, each answering a different question.
The three pillars
- Metrics are numeric measurements aggregated over time, like request rate or error count. They answer is something wrong cheaply and are great for dashboards and alerts.
- Logs are timestamped records of discrete events. They answer what exactly happened with rich detail at a single point.
- Traces follow one request across services. They answer where the time went in a distributed flow.
They work together
A typical investigation flows across all three: a metric alert fires, a trace shows which service is slow, and logs from that service reveal the exact error.
No single pillar is enough. Metrics lack detail, logs lack the cross service view, and traces lack aggregated trends.
Key idea
Observability stands on metrics, logs, and traces, each answering a distinct question that the others cannot.