Monitoring and Observability

Concepts, tooling, and operational practices for monitoring production systems and responding to incidents effectively.

1. The Three Pillars of Observability

Logs

• Discrete events emitted by applications (errors, state changes, audit trails).
• High cardinality: every event can carry unique context.
• Best for debugging specific incidents after the fact.

Metrics

• Numeric measurements aggregated over time (counters, gauges, histograms).
• Low cardinality by design: labels should have bounded value sets.
• Best for detecting trends, setting thresholds, and alerting.