Monitoring

Watching known signals so you learn about problems before your customers do.

Quick Definition

Monitoring is the collection and tracking of predefined metrics and logs to know whether systems are healthy and performing as expected. It answers known questions about system state and underpins alerting, capacity planning, and reliability, a foundation that observability builds upon.

Monitoring is the continuous collection and evaluation of known signals from systems: CPU, latency, error rates, queue depths, and the alerts that fire when they cross thresholds. It answers the question you knew to ask in advance, which distinguishes it from observability, the ability to investigate questions you did not anticipate.

Two cost angles matter. First, monitoring platforms themselves are now major budget items; per-host, per-metric, and per-log pricing scales fast, and log volume in particular can run away. Second, monitoring should include spend: budget alerts and cost anomaly detection are monitoring for money, catching the runaway job with the same urgency as the failing server.

Example. A team adds cost monitoring beside its uptime dashboards: daily spend per service with alerts at 30 percent deviation. The next misconfigured deployment is caught in four hours instead of appearing on an invoice three weeks later.

Audit your own monitoring bill regularly: unused dashboards, unwatched metrics, and over-retained data accumulate like any other waste. The 19 application monitoring tools roundup and the Datadog pricing guide map the landscape.