Intro: Prometheus
summary
Prometheus explains its actual internals: numeric time series (counters vs gauges), a pull model with service discovery, two label sources, and a vector-style query language that enables math across metrics. Benchmarks show 200,000 samples/s per CPU and about 1.3 bytes per sample typical compression. Operational advice covers recording rules to cut memory, pairing for HA, long‑term options like Cortex and Thanos, and deployment notes: running inside Kubernetes is normal but external monitors are needed for cluster-availability checks.
AI-written summary · about
chapters (14)
- 0:00Start
- 0:22Who are we?
- 1:07Prometheus 101
- 3:04Main selling points
- 3:21Working assumptions & concepts
- 4:21Time series
- 5:27Efficiency
- 6:36Exposition format
- 9:14Toil
- 10:33Sanity & sleep
- 12:25Perspective & Incentives
- 14:01Post-Mortems
- 15:33Leverage
- 17:09Oracle
