Which of the following signals belongs to symptom-based alerting?

Symptom-based alertingfocuses on detectinguser-visible or service-impacting issuesrather than internal resource states. Metrics likeAPI latency,error rates, andavailabilitydirectly indicate degraded user experience and are therefore the preferred triggers for alerts.

In contrast,resource-based alerts(like CPU usage or disk space) often represent underlying causes, not symptoms. Alerting on them can produce noise and distract from actual service health problems.

For example, high API latency (http_request_duration_seconds) clearly reflects that users are experiencing delays, which is actionable and business-relevant.

This concept aligns with theRED(Rate, Errors, Duration) andUSE(Utilization, Saturation, Errors) monitoring models promoted in Prometheus and SRE best practices.