Technology · Ebook

Thanos: Highly Available, Long-Term Prometheus

by Shriira Press

Thanos is the CNCF project that extends Prometheus into a highly available, long-term, globally-queryable metrics platform — adding a global query view across all your Prometheus servers, unlimited long-term retention via cheap object storage, HA with deduplication, and downsampling, all Prometheus-compatible. This free book teaches it from the ground up: the Prometheus scaling problem and what Thanos is, Prometheus and metrics concepts, Thanos's architecture (the components and two models), the Sidecar and global query (the foundation, external labels), long-term storage (object storage, the Store Gateway), the Compactor and downsampling, high availability and deduplication, the Receiver and push-based ingestion, querying, rules, and operations, and using Thanos in practice. Ten focused chapters with clear diagrams that make scaling Prometheus concrete — keep your Prometheus servers and add Thanos to get a unified global view across clusters, affordable years-long retention, and reliable HA monitoring, incrementally and without relearning (same PromQL and Grafana).

Contents

1. 1Preface
2. 2Chapter 1 — What Thanos Is
3. 3Chapter 2 — Prometheus and Metrics
4. 4Chapter 3 — Thanos Architecture
5. 5Chapter 4 — The Sidecar and Global Query
6. 6Chapter 5 — Long-Term Storage
7. 7Chapter 6 — The Compactor and Downsampling
8. 8Chapter 7 — High Availability and Deduplication
9. 9Chapter 8 — The Receiver and Push-Based Ingestion
10. 10Chapter 9 — Querying, Rules, and Operations
11. 11Chapter 10 — Thanos in Practice