DuckDB
In-process analytical SQL database for fast local data work.
About
DuckDB is an in-process analytical database — often described as SQLite for analytics — that runs embedded in your application with no server to manage. It speaks a friendly PostgreSQL-inspired SQL dialect and is optimized for OLAP workloads: columnar vectorized execution, parallel query processing, and direct querying of CSV, JSON, Parquet, and Iceberg files locally or in object storage without loading them first.
The project is MIT-licensed and governed by the independent DuckDB Foundation, with 42,000+ GitHub stars and native clients for Python, R, Java, Go, Node.js, Rust, and ODBC. Extensions add spatial queries, full-text search, and more, while newer ecosystem pieces like DuckLake (a SQL-native lakehouse format) and Quack (a client-server protocol) push DuckDB from embedded analytics into shared data-platform territory. It is the default choice for local data pipelines, notebook analytics, and test fixtures that need real SQL.
What it can do
- In-process OLAP SQL engine with vectorized execution
- Direct queries over CSV, JSON, Parquet, and Iceberg files
- PostgreSQL-inspired SQL with rich analytics functions
- Extension ecosystem: spatial, FTS, HTTP, cloud storage
- Native clients for Python, R, Java, Go, Node.js, Rust, ODBC
Setup
Install per platform: pip install duckdb, curl https://install.duckdb.org | bash, or npm install @duckdb/node-api.
Prerequisites
None — zero dependencies, runs in-process.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: Verified via official site at https://duckdb.org (MIT license, DuckDB Foundation governance, format support, client list); GitHub duckdb/duckdb: 42,180 stars via API 2026-10-15; DuckLake/Quack ecosystem details from official site. License: MIT.
★ 42,180 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.