chdb-io/chdb

Health v2 · maintenance only

Adoption not part of the score

2879 stars · 129 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

chDB is an in-process SQL OLAP engine powered by ClickHouse, embeddable in Python (plus Go, Rust, Node.js, and Bun bindings) without running a ClickHouse server. It supports 70+ data formats like Parquet, CSV, JSON, and Arrow, and offers a pandas-compatible DataStore API that compiles to optimized ClickHouse SQL.

Use cases

• run fast OLAP SQL queries on local files without installing a database server
• query parquet and csv files with SQL from python
• speed up pandas groupby and filter pipelines on millions of rows
• embed an analytics database inside an application with zero dependencies
• analyze large datasets in a notebook with ClickHouse performance
• exchange data zero-copy between dataframes and a SQL engine
• prototype analytical queries before deploying to a ClickHouse cluster

When to choose

• you need ClickHouse-style analytical query performance embedded in a Python, Go, Rust, or Node.js app
• you want to query local Parquet, CSV, JSON, or Arrow files with SQL without a server
• your pandas workflows are too slow and you want a drop-in faster API
• you need a zero-dependency embedded OLAP engine for local development or testing

When to avoid

• you need a multi-user production database server with distributed scaling
• you require OLTP-style transactional workloads with frequent small writes
• your platform is Windows-only, since official Python support covers macOS and Linux
• you need heavy concurrent write access from multiple processes

Facets

library · maturity active

database database-driver data-science cli databases data-science analytics big-data developer-tools python go rust bun cli olap embedded-database clickhouse sql in-process pandas dataframe parquet columnar-database macos linux nodejs

Member repositories

For agents

markdown · JSON · MCP: product_card(name="chdb-io/chdb")