Columnar DBMS: The Magic of Data Compression and Super-Fast Queries
summary
Columnar databases, their storage and execution model, and why usage patterns lead to different physical data layouts are explained. The talk covers column compression and vectorized computation, practical challenges of supporting ACID transactions in column stores, and compares performance using the ClickBench benchmark for disk-efficiency measurements.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 1:59What is DBMS
- 3:56Relational data models
- 5:35DBMS workload types
- 7:57DBMS memory hierarchy
- 8:18Data storage models
- 9:53NSM storage model
- 11:30NSM analytics issues
- 12:36DSM storage model
- 14:18PAX hybrid model
- 16:42Columnar DBMS compression
- 18:31Data compression algorithms
- 23:51Data skipping and optimization
- 26:28Query execution engine
- 29:54Vectorized engines
- 30:58Columnar DBMS issues
- 36:05Benchmarking problems
- 38:25Popular benchmarks
- 42:41ClickBench examples
- 50:39Q&A
related talks
26mColumnar Storage: Redefining Data Management for the Modern Era
42mColumnar Storage: Redefining Data Management for the Modern Era
45mClickHouse: what is behind the fastest columnar database
20mColumnstore Index Compression in a Nutshell
52mImplementing OLAP: how we built columnar storage in YDB
34mClickHouse® - What is Behind the Fastest Columnar Database?- 20mColumnstore Index Compression in a Nutshell
18mBuilding Data Lake with MariaDB ColumnStore