A History of Automatic Aggregations

Learn how Criteo developed and evolved their automatic aggregations system to manage hundreds of millions of active metrics in this 19-minute conference talk from the Cloud Native Computing Foundation. Discover how automatic aggregation works as a system of recording rules that matches metrics and removes certain dimensions like instance identifiers to reduce cardinality, making queries faster and significantly reducing metrics storage costs. Explore Criteo's journey from using this approach as a workaround to making it a cornerstone of their backend stability and reliability strategy at scale. Understand how their "zero-effort Observability" philosophy eliminates the need for teams to write or maintain recording rules manually. Follow the evolution of their implementation approaches, examine the lessons learned from different iterations, and see how their latest version leverages recent Prometheus improvements including the new type label feature to integrate seamlessly into their Prometheus-based monitoring stack.