DuckDB is a free, open-source embedded analytical database built and maintained by the DuckDB Foundation and a large community of contributors — often described as "SQLite for analytics." It ranks #21 in real-world database usage at roughly 3.3% adoption, and it keeps growing because it runs directly inside a Python or R process with zero server to set up, making it a favorite for fast local data analysis.
DuckDB's signature trick: you can run SQL directly against a file on disk — no import step, no schema setup — and it's optimized to make that fast even on large Parquet files.
SELECT region, SUM(revenue) AS total_revenue
FROM 'sales_2026.parquet'
WHERE order_date >= '2026-01-01'
GROUP BY region
ORDER BY total_revenue DESC; -- reads the Parquet file directly, no loading stepDuckDB installs as a single library — most people reach it through Python or R, or the standalone CLI for quick exploration.
# install the Python package
pip install duckdb
# or grab the standalone CLI and query a file immediately
duckdb
D SELECT * FROM 'data.csv' LIMIT 10;