Metric

A metric is a numeric measurement of some property of a system, recorded over time. Metrics are one of the core forms of telemetry and are the basis for most dashboards and alerts.

How It Works

Metrics are typically stored as time series: a sequence of values, each with a timestamp and a set of labels that identify the source, such as service, instance, or endpoint. Common metric types include counters, which only increase and track totals like requests served; gauges, which can rise and fall and capture current values like memory in use; and histograms, which bucket observations to summarize distributions such as request latency. Because metrics are aggregated rather than recording every event, they are cheap to store and fast to query, which makes them ideal for alerting.

Why It Matters

Metrics give a continuous, low-cost view of system health and behavior. They power alerting rules, capacity planning, and the service level indicators used to track reliability targets. Their main limitation is that aggregation discards detail; a metric can tell you that error rate rose but not which specific requests failed. For that, engineers turn to logs and traces. Effective observability combines metrics with those richer signals.

Related Terms

Metrics are a form of telemetry central to observability. They are used to compute service level indicators and complement structured logging and tracing.