chDB

active

by ClickHouse, Inc. · since 2023

Embedded, in-process OLAP engine powered by ClickHouse, used mainly from Python.

columnar / olaprelationaldocumentdatabase engineopen source (permissive)emerging

Overview

chDB packages the ClickHouse engine as an in-process library, so a Python (or Go, Rust, Node, C) program can run ClickHouse SQL over Parquet, CSV, Arrow, pandas DataFrames, S3 or remote databases without a server. Sessions can persist tables to a local directory, and newer releases add a pandas-compatible DataStore API. It joined ClickHouse, Inc. in 2024. chDB gives you ClickHouse's function library and formats in a DuckDB-like footprint; it is single-process, the wheel is large, and its non-Python bindings are less mature.

Strengths & trade-offs

Strengths

  • +Full ClickHouse SQL and function library inside a Python process
  • +Reads 70+ formats including Parquet, Arrow and pandas DataFrames zero-copy
  • +Queries S3, MySQL, Postgres and ClickHouse servers via table functions
  • +Backed by ClickHouse, Inc. and tracks upstream releases

Trade-offs

  • −Large binary/wheel size
  • −Single-process, no concurrent multi-user access
  • −Bindings outside Python are less mature
  • −Smaller community than DuckDB

Great for

  • +Notebook analytics with ClickHouse SQL
  • +Local ETL on Parquet and CSV
  • +Testing ClickHouse queries without a server
  • +Embedding analytics in Python services

Think twice for

  • −Multi-user server workloads
  • −Windows-first environments
  • −OLTP

Capabilities

Native or official first-party support · 7 of 25 tracked

✓Joins
✓Geospatial
✓JSON
✓Columnar storage
✓In-memory
✓Durable persistence
✓Compression
–Transactions
–Secondary indexes
–Full-text search
–Vector search
–Change data capture
–Triggers
–Stored procedures
–Materialized views
–TTL / expiry
–Pub/Sub
–Encryption at rest
–Role-based access
–Time travel
–Branching
–Online schema changes
–Streaming ingest
–Offline sync
–Runs in WebAssembly

Scale & consistency

Comfortable data size
Medium · ≤ 10 TB
Horizontal scaling
None
Multi-region
No
Transactions
None
Consistency
Strong
Schema
Strict
Architecture
Embedded, In memory, Single node
Query languages
SQL (ClickHouse dialect), Python DataFrame API

Single process using all local cores; bound by one machine.

Deployment

Self-hosted Vendor-managed cloud Serverless Embedded Edge Kubernetes operator

Ecosystem

compatible withClickHouse
apis & protocolsPython API, C API, DB-API 2.0
client languagesPython, Go, Rust, JavaScript, C, Bun

Alternatives to chDB

Compare all→
DuckDBColumnar / OLAP63.7

In-process columnar SQL database: the SQLite of analytics, running inside your app or notebook.

chdb vs duckdb →
ClickHouseColumnar / OLAP69.8

Open-source columnar OLAP database known for very fast aggregations over billions of rows.

chdb vs clickhouse →
Apache DataFusionColumnar / OLAP31.6

Extensible Rust query engine library on Apache Arrow for building databases and data systems.

chdb vs apache-datafusion →
AlaSQLRelational20.2

Pure-JavaScript in-memory SQL engine for browsers and Node.js that queries arrays, JSON, CSV and Excel files.

chdb vs alasql →
MotherDuckColumnar / OLAP37.7

Serverless cloud data warehouse built on DuckDB, with hybrid local-plus-cloud execution.

chdb vs motherduck →
Apache Spark SQLRelational50.3

Spark's SQL and DataFrame engine for large-scale batch, streaming and lakehouse processing.

chdb vs apache-spark-sql →

chDB FAQ

What is chDB?

chDB is a columnar / OLAP database by ClickHouse, Inc., first released in 2023. chDB packages the ClickHouse engine as an in-process library, so a Python (or Go, Rust, Node, C) program can run ClickHouse SQL over Parquet, CSV, Arrow, pandas DataFrames, S3 or remote databases without a server. Sessions can persist tables to a local directory, and newer releases add a pandas-compatible DataStore API. It joined ClickHouse, Inc. in 2024. chDB gives you ClickHouse's function library and formats in a DuckDB-like footprint; it is single-process, the wheel is large, and its non-Python bindings are less mature.

Is chDB open source?

Yes. chDB is open source. License: Apache License 2.0.

Does chDB support ACID transactions?

No. chDB does not provide ACID transactions; its consistency model is strong.

Does chDB scale horizontally?

chDB does not scale horizontally; it runs on a single node, so you scale it up rather than out. It is comfortable at medium scale (≤ 10 TB).

How can I run chDB?

You can self-host it and embed it in your application.

What is chDB best for?

chDB is a strong fit for notebook analytics with ClickHouse SQL; local ETL on Parquet and CSV; and testing ClickHouse queries without a server. Think twice for multi-user server workloads.

What are the best alternatives to chDB?

Common alternatives to chDB include DuckDB, ClickHouse, Apache DataFusion, AlaSQL and MotherDuck. The right choice depends on your workload, scale and hosting; WhichDB's smart match ranks them against your requirements.

How popular is chDB?

chDB ranks #143 of 844 databases on the WhichDB Score (21.1/100), and #14 among columnar / OLAP databases. The score combines public adoption signals such as package downloads, GitHub activity, job posts and developer discussion.