Source: ClickHouse
“Column-oriented databases are better suited to OLAP scenarios. They are at least 100x faster in processing most queries,” says the ClickHouse home page, and it explains why in two paragraphs that are better than most textbook introductions: a row store keeps rows together, a column store keeps columns together, and an analytics query usually needs only a few columns. Further down there’s ClickHouse Local, which runs SQL on CSV and Parquet files without a server.
TL;DR
- Open-source column-oriented OLAP (online analytical processing) database, SQL
- Vendor claim: column stores “at least 100x faster” on most queries
- FAQ credits compression and vectorized query execution
- Often a speed layer on top of a data warehouse or OLTP (transaction) database
- Self-managed: free;
curl https://clickhouse.com/ | sh, Docker, Windows, or playground - Cloud: AWS, GCP, Azure; from $50/month; 30-day trial, $300 credits
- ClickHouse Local: queries CSV, TSV, Parquet files, no server
- chDB: in-process SQL engine on ClickHouse, Pandas-compatible API
- Scale: 3.3k+ contributors, 894+ releases, 50.2k+ GitHub stars
- Customers: Sony, Lyft, Cisco, GitLab; quotes from Anthropic, Tesla, Lyft