Alerting
Alerting notifies on-call staff when metrics cross thresholds or anomalous events occur, spanning alert rules, routing, and notification channels.
Purpose
Detect anomalies promptly and notify for handling, shortening time-to-detection.
Mechanism
Rules: based on metric thresholds (CPU>90%), ratios (error rate>1%), and anomalous events. Routing: dispatched by severity/service. Notification: PagerDuty/Slack/email. Inhibition and grouping prevent alert storms.
Use cases
- Incident detection
- On-call notification
- Monitoring