Knowledge MaxxingObservability
Observability is how well you can understand a system's internal state and diagnose problems just from its external outputs, like logs, metrics, and traces.
Prerequisite: Trace
What matters
A highly observable system lets engineers answer new, unanticipated questions about a problem using existing data, rather than needing to add new instrumentation after every incident.
Example
A well-instrumented system might let an engineer diagnose an unexpected slowdown using existing logs, metrics, and traces, without deploying any new code to investigate.
Common misconception
Observability is not the same as having a lot of logs; it specifically means answering new questions about the system's behavior from data already being collected.
Recall question
What distinguishes a genuinely observable system from one that just has a lot of logs?