Metrics are numeric values sampled or counted at regular intervals and stored as time series. Typical examples are requests per second, error rate, CPU utilization, queue depth and response time.
Why it matters
Metrics are compact and cheap to store, which makes them the best signal for dashboards and alerts. A single series can cover months of history, so you can compare today with last week and see slow trends such as a memory leak. They tell you that something changed, though usually not why, which is where logs and traces help.
Example
Four common metric types:
- Counter: only goes up, such as
http_requests_total. - Gauge: goes up and down, such as
memory_used_bytes. - Histogram: counts values in buckets, used to derive percentiles of latency.
- Rate: a counter converted to change per second.
An alert might fire when the 5xx error rate stays above 2% for five minutes.
Common mistakes
- Using high-cardinality labels such as user IDs. Every unique combination creates a new series and can overwhelm storage.
- Averaging latency instead of tracking percentiles.
- Alerting on causes (high CPU) rather than symptoms users feel (errors, slow pages).
- Changing metric names without updating dashboards and alert rules.
Related
Get notified when this term updates
Plus occasional practical tips on uptime and monitoring, nothing else. Unsubscribe anytime.
This page is general information written by our editorial team and updated when documentation changes. Servers and providers differ, so test changes safely first. See our editorial policy and disclaimer.