The Three Pillars of Observability — Metrics, Logs, Traces

Why it matters

Wrong pillar for the question wastes debugging time. Understanding which reveals what makes ops fast.

The architecture

Metrics: numeric measurements sampled at intervals. CPU %, request rate, error count. High cardinality expensive.

Logs: text records of events. Rich context, expensive at scale.

Traces: request paths across services. Show causality.

How it works end to end

Metrics answer 'what is happening now' (dashboards, alerts).

Logs answer 'what happened' (search, debugging).

Traces answer 'why is this slow' (causality, service dependencies).

Combined: metric alerts → find trace → drill to log for full context.