Long-Term Metrics with M3 and Prometheus
summary
Explains how Prometheus uses remote_write and remote_read to offload storage to m3, showing a real m3db deployment: a 9-node cluster across 3 zones with RF=3 and quorum=2 and a stateless coordinator that fans out reads and writes, rebalances via peer streaming when nodes are added, and has no built-in autoscaling. Demonstrates two Prometheus instances remote-writing into m3 and querying a unified view, claims 100% PromQL compatibility by embedding the Prometheus engine, and notes tradeoffs: self‑managed disks, backups, and capacity planning.
AI-written summary · about
chapters (8)
- 0:00Intro
- 2:31Single Instance Challenges
- 8:44Prometheus Remote Write
- 11:43High Availability
- 16:50Data Rebalancing
- 17:58Prometheus Remote Read
- 20:32M3: Prometheus Compatibility
- 25:46Comparing M3 with Other Rernote Storage
related talks
37mM3 and Prometheus, Monitoring at Planet Scale for Everyone
11mAggregating, Alerting and Graphing on Millions of Prometheus T...
26mWhen Prometheus Can’t Take the Load Anymore
31mWhen Prometheus Can’t Take the Load Anymore
05mMigrating Prometheus data between different storage systems
42mPrometheus Intro and Deep Dive
29mStreaming Recording Rules for Prometheus, Thanos, and Cortex Using...
33mMonitoring Ceph with Prometheus