Metric
A metric is a quantifiable time-series value (counter/gauge/histogram/summary) that monitors system health and performance.
Purpose
Quantify system behavior, driving alerting, capacity planning, and SLOs.
Mechanism
Counter: monotonically increasing (request counts). Gauge: rises and falls (memory/connections). Histogram: distribution buckets (latency distribution). Summary: quantiles (p99 latency). Prometheus exposes /metrics.
Use cases
- Monitoring
- Alerting
- SLO