Data Vault

Data vault is a modeling methodology for the warehouse integration layer that splits every entity into three parts: hubs (just the business keys), links (relationships between keys), and satellites (descriptive attributes with full history). Everything is append-only and stamped with load time and source, so the vault absorbs new sources and schema change by adding tables, never altering them — and history is auditable by construction.

The cost is table count and joins: a vault is not meant to be queried by humans, so a presentation layer of star schemas is built on top. It earns its complexity in large, regulated, many-source estates; smaller shops usually go straight to dimensional.

Go deeper: Data Vault vs Dimensional Modeling: Which Belongs Where