Amundsen

Open-source data discovery and metadata catalog from Lyft

AI Data AnalysisFree planOverseasβ˜…β˜…β˜…β˜†β˜† 3.0

What is Amundsen?

Amundsen is an open-source data discovery and metadata platform originally built at Lyft. It indexes tables, columns, dashboards and owners into a searchable catalog, and shows popularity and usage data alongside each asset so people can find the tables their colleagues actually rely on. Lineage, column descriptions and ownership records are displayed in one page, and the metadata layer is modular so individual components can be adopted on their own. It is worth knowing that development has slowed considerably in recent years, so teams evaluating it should weigh the active community and available maintenance against its solid core design.

Last updated: 2026-09-20. This site only provides an index; for exact features, pricing, and licensing, see the official website.

Key features

  • Searchable catalog of tables, columns, dashboards and owners
  • Usage and popularity statistics to rank which tables matter
  • Table and column level lineage views for impact analysis
  • Modular services that can be adopted individually rather than wholesale
  • Python and REST APIs for pushing in custom metadata
  • Pluggable metadata stores, including graph and search backends

Pros & cons

Strengths

  • Free and open source with no per-seat licensing
  • Usage-based ranking helps newcomers find the right table quickly
  • Modular design allows a staged adoption

Watch out for

  • Development activity has slowed, so evaluate long-term maintenance
  • Initial deployment and metadata ingestion take real effort
  • Features trail commercial catalogs in automation and governance

Best for & use cases

data discovery, metadata cataloging, lineage exploration and self-hosted governance

If you're comparing similar products, check the alternatives below, or browse all tools in the AI Data Analysis category.

FAQ

Is Amundsen still actively developed?

Community activity has decreased noticeably compared with its early years, and much of the ecosystem has moved on. It still works and can be self-hosted, but plan for in-house maintenance capability or evaluate more actively maintained alternatives.

What do I need to run it?

A metadata database, a search backend such as Elasticsearch, a graph store for lineage, and the web and metadata services, usually containerised. Expect to write ingestion jobs for the metadata sources you care about.

What problem does a catalog solve?

It answers which table to use, who owns it and what depends on it. Without that, analysts rebuild existing datasets and break downstream jobs unknowingly. That is a human problem first and a tooling problem second.