ClickHouse as Time-Series Storage for Graphite
summary
ClickHouse serves as the time-series store for Graphite with Yandex’s open-source Graphouse acting as the bridge: it replaces Carbon, batches inserts into ClickHouse, and keeps a metric tree with top 3 levels in memory and deeper levels loaded on demand. Using ClickHouse’s Graphite MergeTree for on-merge compaction, the setup sustains >2M metrics/sec in Yandex’s inserts benchmark (go‑carbon+whisper reaches ~1M, BigGraphite/Cassandra underperform). Planned work converts Graphite functions into ClickHouse SQL to push aggregation into the database and remove graphite-web bottlenecks.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:51Graphite overload and ClickHouse
- 3:58ClickHouse time-series design
- 5:45Graphite Merge table engine
- 6:06GraphHouse replacing Carbon/Whisper
- 8:57GraphHouse with ClickHouse cluster
- 10:25ClickHouse hardware/sharding
- 11:26Benchmarking time-series stores
- 13:45Graphite workflow bottlenecks
- 14:52Offloading to ClickHouse
- 17:01Optimizing ClickHouse/GraphHouse
- 18:49Future benchmarks/optimizations
- 20:21Grafana with ClickHouse
related talks
55mYandex.Metrica and a Non‑standard ClickHouse
26mO11y in One: Exploring ClickHouse as a Unified Telemetry Database
42mClickHouse: what is behind the fastest columnar database
44mUnconventional Uses of ClickHouse
31mHow ClickHouse® Inspired Us to Build a High Performance Time Series Database
45mClickHouse: what is behind the fastest columnar database
51mWhat You Need to Know About ClickHouse Architecture
53mClickHouse capabilities for advanced developers