• Comment: Please rewrite from scratch without the use of LLMs. Corporate blogs are not reliable sources. Helpful Raccoon (talk) 23:23, 3 June 2026 (UTC)

chDB
Original authorAuxten Wang
DevelopersClickHouse, Inc.
Release2023
Written inC++, Python
Operating systemmacOS, Linux
TypeEmbedded database, OLAP
LicenseApache License 2.0
Websiteclickhouse.com/docs/chdb
Repositorygithub.com/chdb-io/chdb

chDB is an open-source embedded SQL engine for online analytical processing (OLAP), based on ClickHouse. It runs the ClickHouse query engine in-process, inside a host application, rather than through a separate database server, and is distributed mainly as a Python package under the Apache License 2.0.[1][2] In addition to Python, the project provides bindings for C, C++, Go, JavaScript, Bun, and Rust.[3]

History

edit

chDB was created by Auxten Wang and first published as a Python package in 2023, with version 1.0.0 following in December of that year.[4][5][6] It was acquired by ClickHouse, Inc. in March 2024.[4][2]

Design

edit

Because the ClickHouse engine runs in the same process as the calling application, data moves between them without being copied, so a query can read values already held in memory, such as a pandas DataFrame in a Jupyter notebook, and return its results into the same session.[1]

chDB uses the same SQL as ClickHouse and can read the formats ClickHouse supports, more than 70 in total, among them Parquet, CSV, and JSON.[7][8][9] A single query can combine these with in-memory data and remote sources without loading them into a server first, and a session can keep data between queries for multi-step work.[9] Its Python API works directly with dataframes, which makes it convenient to call from existing data-science code.[10]

Like DuckDB, chDB is an embedded OLAP database, and the two are often compared for local analytics.[11][12] It is mainly an option for developers already working with ClickHouse, especially when a workload relies on ClickHouse-specific SQL features such as array functions.[11][12]

References

edit
  1. 1 2 Schulze, Robert; Schreiber, Tom; Yatsishin, Ilya; Dahimene, Ryadh; Milovidov, Alexey (2024). "ClickHouse - Lightning Fast Analytics for Everyone" (PDF). Proceedings of the VLDB Endowment. 17 (12): 3731–3744. doi:10.14778/3685800.3685802.
  2. 1 2 "chDB". Database of Databases. Carnegie Mellon University Database Group. 14 March 2024. Retrieved 2026-06-04.
  3. "chdb-io/chdb". GitHub. chdb-io. Retrieved 2026-06-04.
  4. 1 2 Milovidov, Alexey; Bragin, Tanya (2024-03-06). "chDB joins the ClickHouse family". ClickHouse Blog. ClickHouse, Inc. Retrieved 2026-06-04.
  5. "chdb 0.2.2". Python Package Index. 2023-03-23. Retrieved 2026-06-04.
  6. "Release v1.0.0". GitHub. chdb-io. 2023-12-01. Retrieved 2026-06-04.
  7. Archer, Cameron (2025-10-27). "ClickHouse vs chDB: let's talk about embedded ClickHouse". Tinybird. Retrieved 2026-06-04.
  8. Orlov, Dmitrii (2024). Knowledge Graphs at Your Fingertips: a Survey of Kaggle Notebooks for Knowledge Graph Analytics (PDF) (Master's thesis). Eindhoven University of Technology.
  9. 1 2 Zhiyanov, Anton (2023-12-11). "Trying chDB, an embeddable ClickHouse engine". Anton Zhiyanov's blog. Retrieved 2026-06-04.
  10. Wang, Xiaoying; Wang, Jiannan; Wang, Tianzheng; Zhang, Yong (2025). "Accio: Bolt-on Query Federation" (PDF). Proceedings of the VLDB Endowment. 18 (7): 2126–2135. doi:10.14778/3734839.3734849.
  11. 1 2 Krupitsky, Daniel (2025-03-25). "Small-Scale Data Pipeline: DuckDB vs. chDB vs. Spark". Digital Turbine. Retrieved 2026-06-04.
  12. 1 2 Roset, Martin-Pierre (2026-04-15). "Embedded Databases in 2026: DuckDB, SQLite, Polars, and chDB". Kestra. Retrieved 2026-06-04.