Keynote: Prometheus: From Berlin to Bonanza
summary
Prometheus traces from a 2012 Berlin project to a CNCF staple with a time series DB, local storage, and a powerful query language PromQL that lets you predict disk fill and compute 95th percentile latency. It auto-discovers targets from Kubernetes, exposes metrics via client libraries and exporters, and routes alerts through Alertmanager with grouping. The project reports ~300 core contributors, ~150 integrations and an estimated 500–1,000 production users. Ongoing work includes long‑term storage and a 20–30% memory reduction in the next release.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 0:25History
- 2:48Prometheus
- 3:33Philosophy
- 4:30Discovery
- 5:07Libraries
- 6:49Query Language
- 8:44Grouping
- 9:58Monitoring
