Open Data Infrastructure
Data Catalog vs Metastore vs Business Glossary
A practical distinction between data catalogs, metastores, and business glossaries, and why the catalog is becoming ODI control plane.
Data catalog, metastore, and business glossary get used like synonyms. They are not synonyms, and that confusion makes architecture decisions worse.
Set clear definitions
A metastore is a technical metadata store. Historically, it helped engines find tables, schemas, partitions, and storage locations. A business glossary defines business terms, owners, metrics, and shared language. A data catalog can include both, but its bigger job is discovery, governance, access, and operational coordination.
Those boundaries blur in real products. That is fine. The architecture still needs to know which responsibility lives where.
The catalog is becoming the control plane
In an open lakehouse, the catalog increasingly coordinates table operations, permissions, discovery, identifiers, and metadata access. That makes it more than a search box. It becomes part of the control plane.
The glossary still matters because humans and agents need business meaning. The metastore still matters because engines need technical facts. The catalog matters because it is where those facts start to become enforceable infrastructure.
Core idea: a glossary explains meaning, a metastore tracks technical table facts, and a catalog turns metadata into a usable control surface.
AI makes the distinction more important
Agents do not just need table locations. They need allowed data, trusted definitions, lineage, freshness, ownership, and quality signals. If those live in disconnected systems, the agent either guesses or gets over-permissioned access. Both are bad.
A good ODI catalog strategy connects technical metadata with governance and semantic context without pretending one UI solves every problem.
Questions to ask
- Which system is authoritative for table operations?
- Which system is authoritative for business definitions?
- Which system enforces or exposes access decisions?
- Which metadata can be used by engines, services, and agents?
If those answers are fuzzy, the architecture is fuzzy too.
Sources to start with
These are the primary sources I would start from when checking the claims in this piece.