On-call

On-call is a practice in which designated engineers are available outside normal working hours to respond to incidents affecting a production system. When alerts fire – indicating degraded service, errors, or outages – the on-call engineer is responsible for investigating, triaging, and either resolving the issue or escalating it. The practice is a core component of site reliability engineering and DevOps, ensuring that someone is always accountable for system health.

An on-call rotation distributes this responsibility across a team so that no individual is perpetually burdened. Rotations may follow various schedules – weekly, biweekly, or split into primary and secondary tiers – and typically distinguish between weekday and weekend coverage. The goal is to balance rapid response times against the risk of alert fatigue, where excessive or noisy alerts desensitize engineers and degrade the quality of responses.

Effective on-call depends on good observability. Engineers need logs, traces, metrics, and runbooks to diagnose problems quickly under pressure. Conversely, on-call experience feeds back into the system: recurring incidents expose weak points in the architecture, and the insights gained should drive improvements that reduce future alert volume.

On-call is one of many conventions that make up a team’s ways of working. It intersects with alerting strategy, incident management processes, and post-incident review practices.