databend
Data Agent Ready Warehouse : One for Analytics, Search, AI, Python Sandbox. — rebuilt from scratch. Unified architecture on your S3.
What is databend?
Data Agent Ready Warehouse : One for Analytics, Search, AI, Python Sandbox. — rebuilt from scratch. Unified architecture on your S3.
Overview of databend
What it is
Databend is an open-source data warehouse built in Rust. It is designed to handle analytics, search, and agent workloads on the same tables, eliminating the need for a second system to sync. The warehouse runs on object storage such as S3, Azure, and GCS, and supports SQL analytics, vector search, and full-text search within a single system. It also includes sandboxed Python UDFs for agent logic.
Key features
- SQL analytics, vector search, and full-text search in a single warehouse.
- Agent-ready architecture with sandboxed Python UDFs for agent logic and SQL for orchestration.
- Elastic compute on object storage (S3, Azure, GCS) for scalability.
- Git-like data versioning (branching) for safe experimentation on production snapshots.
- Auto schema evolution and transaction support.
- Published benchmark results: TPC-H SF100 vs. Snowflake shows 1.25× faster and ~60% lower cost; data ingestion vs. Snowflake shows 5.15× faster and 91% lower cost; vector and full-text search on 10M docs shows vector p50 41 ms and full-text p50 35 ms.
Use cases
- AI Agents: Use sandbox UDFs, SQL orchestration, and branching for safe operations.
- Analytics & BI: Large-scale SQL analytics.
- Search & RAG: Vector and full-text search.
Getting started
Cloud (Recommended) — Sign up, create a warehouse, and run SQL in the worksheet. See the Cloud guide.Self-hosted — Run the all-in-one image and connect with BendSQL:bashdocker run -d -p 127.0.0.1:8000:8000 datafuselabs/databendbendsql -u rootSee the Docker image options and self-hosted deployment documentation for credentials, storage, and production setup.
License and pricing
Databend is licensed under Apache 2.0. A small set of enterprise features is under Elastic 2.0. See LICENSING.md and the Licensing FAQ for details.
Based on the project's public README and repository metadata.