symptom-based alerting

Paging on what users feel, such as errors, slowness or unavailability, rather than on internal causes such as high CPU or a restarted process.

In more detail

A cause can change without harming users, and one symptom can have many causes, so cause alerts produce noisy urgency. Keep causes on dashboards and in runbooks as investigation evidence, and page when the customer-facing symptom is urgent.

Related terms

Taught in

References