Alert routing and deduplication

DesignMedium35 min

Your task

Design the alerting and notification routing layer that sits behind a metrics platform for 2,000 engineering teams. Rule evaluators across multiple regions fire ~30k alert state-changes/min during a large outage (normally ~200/min). You must deduplicate identical alerts from redundant evaluators, group related alerts into a single notification, route to the right on-call via escalation policies, and never drop or double-page. Design state management, dedup, grouping/throttling, and delivery guarantees.

It covers

  • Deduplication
  • Grouping
  • Notification delivery