Prometheus from A to Z
summary
Shows a hands-on Prometheus setup on Kubernetes using the Prometheus Operator to deploy Prometheus, Grafana, node_exporter and Alertmanager, and exposes a simple Python app at /metrics with ServiceMonitors for auto-discovery. Demonstrates Prometheus metric types (counters, histograms with histogram_quantile), rate() behavior across resets, and cardinality pitfalls from label explosion; notes storage and performance characteristics (head stores ~2–3 hours, samples compress to ~1–2 bytes, a single server can handle on the order of 10 million series on large machines) and recommends Thanos or Cortex for long-term retention; shows scaling/recovery and alert routing via Alertmanager.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:03Prometheus overview
- 3:33History and adoption
- 7:07Prometheus philosophy
- 14:18Technical operation
- 23:08Basic interface
- 27:11Data model and time series
- 32:33PromQL queries and aggregation
- 44:54Alerts and Alertmanager
- 1:24:11Break
- 1:34:54Application instrumentation
- 1:51:57Monitoring best practices
- 2:03:24Alert management and performance
- 2:12:39Prometheus with Kubernetes
- 2:25:23High availability and scaling
- 2:33:56Conclusion and community
related talks
24mPractical Kubernetes Monitoring with Prometheus
32mDemocratizing Application Monitoring with Prometheus
50mMetrics and alerting with Prometheus and Grafana
16mPrometheus 101
08mPrometheus in 5 minutes
37mMonitoring Kubernetes with Prometheus
30mIntro: Prometheus
26mEnd-to-end monitoring with the Prometheus Operator