Engineering
Observability vs Logging
Why logs are useful, but not enough to understand a running system.
Logs are valuable because they preserve narrative. A well-placed log line can explain a decision, expose an input, or show the exact branch a request took through a service.
Observability is wider than that. It asks whether you can answer new questions about the system without shipping new code. Metrics, traces, structured logs, and domain-specific events each cover a different part of that problem.
The trap is collecting everything and understanding nothing. Useful observability starts from the questions you expect to ask during an incident: where is time going, what changed, who is affected, and which dependency is shaping the failure?