DevOps
What is Metrics?
Numeric time-series measurements — request rates, latencies, error counts — aggregated to track system health and trigger alerts.
Definition
Metrics are numeric measurements sampled over time: request throughput, p99 latency, error rate, queue depth, memory usage. Because they aggregate cheaply, metrics answer "how is the system trending?" far more efficiently than logs. Prometheus popularized the pull model — services expose a /metrics endpoint with counters, gauges, and histograms, and PromQL queries drive Grafana dashboards and alert rules. The RED method (rate, errors, duration) covers services; USE (utilization, saturation, errors) covers resources. Metrics feed SLO tracking, autoscaling decisions, and the alerts that page an on-call engineer.