Install duckdb with Homebrew, apk, MacPorts, Nix, pacman, scoop, winget

Project links

• Homepage
• Repository
• Upstream docs
• Package manager page

Commands and aliases

• duckdb
• duckdb_cli

Project history

DuckDB was created by database researchers Mark Raasveldt and Hannes Muehleisen and grew out of academic work on embedded analytics. The DuckDB site links its research lineage to the SIGMOD 2019 DuckDB demo and the CIDR 2020 embedded analytics paper.

The project deliberately borrowed the embedded, zero-server deployment model made famous by SQLite, but targeted analytical query workloads instead of transactional application storage. Its documentation emphasizes no external runtime dependencies, portability across major operating systems and CPU architectures, and a columnar-vectorized execution engine for OLAP.

Over time DuckDB expanded from a C++ database engine and CLI into a multi-language analytical component with clients for Python, R, Java, Wasm, and other environments. That made it part database, part data-science runtime building block.

Adoption history

Early adoption came from data scientists and engineers who wanted fast local SQL over CSV, Parquet, pandas, Arrow, and data lake files without running a database server. The official README highlights deep integrations with pandas and dplyr, plus standalone CLI usage.

DuckDB's 1.0.0 release in June 2024 was a major adoption milestone because the project used it to signal stability for production use. By then, it had already become a common dependency in notebooks, local analytics scripts, ETL checks, and data toolchains.

Package-manager adoption is broad: the package appears in Homebrew, Linux distributions, Nix, MacPorts, Arch, Scoop, and winget-style Windows packaging. That matters because the CLI is often used directly by people inspecting files at the shell.

How it is used

Typical usage includes ad hoc SQL over CSV and Parquet files, embedding an analytical engine in Python or R, running local transformations before loading data elsewhere, and querying data lake files through extensions such as httpfs and S3 support.

The CLI gives package users a small, scriptable database shell. The library APIs make the same engine available inside applications, notebooks, and data pipelines.

Why package nerds care

DuckDB is one of the clearest modern examples of a package whose CLI understates its ecosystem impact. The Homebrew formula installs a command, but the project is also a database engine, language bindings ecosystem, extension platform, and data tooling dependency.

For package metadata, DuckDB is significant because its category is not just database-server software. It is an embedded OLAP engine whose distribution model looks like SQLite, whose usage overlaps with pandas/Arrow/Parquet workflows, and whose package-manager footprint signals mainstream adoption in local analytics.

Timeline

• 2019: DuckDB: An Embeddable Analytical Database appears as a SIGMOD demo.
• 2020: Data Management for Data Science - Towards Embedded Analytics appears at CIDR.
• 2020: Public 0.1.x releases appear in the GitHub release history.
• 2022: DuckDB-Wasm is presented as a VLDB demo, extending DuckDB into browser-based analytics.
• 2024: DuckDB 1.0.0 is announced as the first stable major release.

Related projects

• SQLite is the central architectural comparison point for DuckDB's embedded model, while MonetDB, PostgreSQL, and vectorized analytical engines are part of its database lineage and testing/reference context.
• DuckDB is closely related in usage to Apache Arrow, pandas, dplyr, Parquet, MotherDuck, and DuckDB-Wasm; these projects and integrations explain much of its adoption outside traditional database administration.

Sources

• https://github.com/duckdb/duckdb
• https://duckdb.org/why_duckdb.html
• https://duckdb.org/2024/06/03/announcing-duckdb-100.html
• https://duckdb.org/library/
• https://github.com/duckdb/duckdb/releases