ClickHouse: what is behind the fastest columnar database
summary
Open source ClickHouse claims throughput of billions of rows and tens of gigabytes per second by using columnar storage with per‑column files and heavy compression, vectorized execution across CPU cores, and a sparse primary index (defaults to every 8192 rows) that lets it skip large blocks. Data lands as parts in MergeTree tables which are merged in background (ReplicatedMergeTree, Replace/Aggregating variants change merge behavior). Integrations include Kafka engine, ClickHouse Kafka Connect, materialized views and external dictionaries. Pitfalls: not ideal for OLTP or single‑row lookups, prefer batched inserts (~10k) and explicit column selection.
AI-written summary · about
chapters (12)
- 0:00Introduction to ClickHouse
- 1:20The challenge of big data
- 6:35OLTP vs OLAP systems
- 9:41ClickHouse architecture
- 13:28Indexing and data skipping
- 17:50Vectorized execution
- 19:06ClickHouse table engines
- 24:23Data ingestion and denormalization
- 27:48Demonstration with historic data
- 31:01Kafka integration
- 34:23Pitfalls and best practices
- 37:26Conclusion and Q&A
related talks
45mClickHouse: what is behind the fastest columnar database
34mClickHouse® - What is Behind the Fastest Columnar Database?
51mWhat You Need to Know About ClickHouse Architecture
28mColumnstore Revolution
59mEffective Use of ClickHouse
40mData warehouse or conventional database: Which is right for you?
26mColumnar Storage: Redefining Data Management for the Modern Era
26mHandling Large Volumes of Immutable Structured Data With ClickHouse