Skip to main content
BRILLIQS

Amundsen

An open source data discovery tool built around search, ranking results by how heavily each asset is used.

Amundsen is an open source data discovery and metadata tool. Its focus is narrow by design: helping people find the data they need. Search is the primary interface, and results are ranked using usage information, so heavily used tables appear ahead of the many that exist but are rarely touched.

A deliberately narrow tool

Amundsen sets out to solve one problem: helping people find the data they need.

Not policy management. Not detailed classification schemes. Not the full apparatus of enterprise governance. Discovery.

That narrowness is a feature rather than a shortcoming, and it is worth understanding why.

Why search rather than browsing

Data platforms are commonly presented as a hierarchy. Systems, then databases, then schemas, then tables.

That structure assumes the person looking knows roughly where to look. Somebody searching for customer subscription data does not know which of four systems holds it, and navigating a tree to find out is slow and dispiriting.

Search matches how the question is actually formed. You know the subject; you do not know the location.

Why usage ranking matters so much

Search alone is not enough, and this is where Amundsen's central idea earns its place.

Search for a common business term in a large organisation and you get forty matches. Most are abandoned experiments, partial copies, tables somebody built for a project three years ago. A handful are the ones people actually rely on.

From names and descriptions alone, they are indistinguishable. That is why search over a large catalogue often fails to help.

Ranking by usage separates them. A table queried hundreds of times a week is almost certainly the one you want. One never queried almost certainly is not.

This works because it uses a signal nobody has to maintain. Documentation decays; usage is regenerated continuously by people doing their work.

Who to ask

Usage information also answers a second question.

When you find a table and need to understand it, someone knows. Usage shows who queries it most, which is a good indicator of who that is.

That converts an open question into a specific one directed at a specific person.

How metadata arrives

Collection is handled by a separate ingestion library that reads source systems and writes into the metadata service.

Keeping ingestion separate means coverage can be extended. A system nobody has written support for can be handled by extending the library, so internal systems appear alongside standard ones.

Architecture

Metadata, search and the interface are separate services, and the storage behind them can be chosen from supported options.

That flexibility matters for teams deploying it themselves, since it allows the deployment to use infrastructure they already run.

Who uses Amundsen

Amundsen is used by data teams that want discovery without adopting a full governance platform, particularly technology organisations comfortable deploying and extending open source.

Points to consider

The narrow scope is the trade. Organisations needing policy management, detailed classification and formal governance workflow will find those absent, and a broader platform is the right answer for them.

Usage ranking requires usage information, so the value depends on ingesting query history from the systems that have it.

It is software you deploy and operate, comprising several services and their storage, which is real infrastructure work.

Getting started

The documentation covers the architecture, deploying the services and using the ingestion library to load metadata from a source. Loading one well used system, including its usage information, demonstrates why the ranking matters more than a bare catalogue would.

Key features of Amundsen

Capabilities described in the official documentation.

Search as the main interface

Finding data is done by searching rather than by navigating a hierarchy of systems and schemas.

Results ranked by usage

How often an asset is queried informs its position in results, so widely used tables surface first.

A separate ingestion library

Metadata is collected by a library that reads source systems and writes into the metadata service.

Service based architecture

Metadata, search and the interface are separate services, which allows the storage behind them to be chosen.

Advantages of Amundsen

Factual advantages that follow from the features above.

The likely answer appears first

Ranking by usage means a search for a common term returns the table people actually use rather than an abandoned copy.

Narrow scope keeps it usable

Because it concentrates on discovery, it stays simple enough that people use it rather than avoid it.

Ingestion can be extended

The collection library can be extended for systems it does not already cover, so coverage is not fixed.

Open source with no licence cost

It can be deployed and evaluated without a purchase decision preceding any practical experience.

Common use cases for Amundsen

Situations the official documentation describes this tool as being used for.

Technology

Finding the right table quickly

An analyst searches for a subject and the heavily used table appears ahead of similar alternatives.

Cross industry

Onboarding a new analyst

Somebody new discovers what data exists by searching rather than by asking colleagues repeatedly.

Data platforms

Identifying who understands a table

Usage information shows who queries an asset most, which indicates who to ask about it.

Platform engineering

Adding coverage for an internal system

The ingestion library is extended so an in house system's assets appear alongside the rest.

Frequently asked questions about Amundsen

Answers taken from the official documentation for this tool.

Because people looking for data have a subject in mind rather than a location. Browsing a hierarchy of systems and schemas assumes you know where to look, which is exactly what somebody searching does not know.

Through a separate ingestion library that reads source systems and writes into the metadata service. Because it is a library, it can be extended to cover systems that are not supported already.

A search for a common term typically matches many similarly named tables, most of which are abandoned or partial. Ranking by how heavily each is queried puts the one people actually use at the top, which is usually the right answer.

No, and deliberately so. Its focus is discovery. Broader governance capabilities such as policy management and detailed classification are the domain of larger platforms, and the narrower scope is part of why it stays approachable.