How We Built ClickStack - an open source O11y stack
summary
ClickStack is an opinionated, open-source observability stack built from ClickHouse (Apache-2) for storage, HyperDX (MIT) for UI, and OpenTelemetry for ingestion, deployable with one command. It sustains ingest rates near tens of GB/s, stores ~100 PB of logs internally using 20x compression ratios and object storage single-copy retention, and converts Lucene-like queries into SQL for safe execution. Query alignment cuts queries from ~1.9s to ~0.015s. PromQL support is landing incrementally; map fields are materialized into columns to avoid expensive full-map reads.
AI-written summary · about
chapters (9)
- 0:00Introduction and Speaker Background
- 0:34Overview of ClickHouse
- 1:52Introduction to ClickStack
- 2:39Choosing the Right Data Store
- 6:17Advantages of Column Stores
- 7:37Optimizing Query Performance
- 12:29OpenTelemetry and Observability
- 13:31Building a User-Friendly Interface
- 20:05Q&A Session
related talks
36mThe Power of Consolidation: A Unified Stack for Business Intelligence, Securi...
27mFast DIY O11y with OpenSrc Analytics & Visualization
18mSwitching Jaeger Distributed Tracing to ClickHouse® to Enable APM
27mDave Stokes talks open - Stocking the flames of open source - Episode 1
34mThe State of SQL-Based Observability
32mSponsored Session: Observability and Agentic AI - From Observability to Answers...
25mUnifying Observability: Weaving Prometheus, Jaeger, and Open Source Together to Win
30mCheap DIY Monitoring with OSS Analytics & Visualization