Data Modelling
Data Vault vs Dimensional Modelling
Two modelling approaches solving different problems — and how to combine them without doubling the work.
6 min read
Different objectives
Dimensional modelling optimises for query comprehension and analytical performance. Data vault optimises for auditable integration of many changing sources, separating business keys, relationships and descriptive attributes so that sources can be added without restructuring existing models.
Choosing between them
The decision usually turns on source volatility and audit requirements rather than data volume.
- Few, stable sources with clear reporting needs: dimensional is usually sufficient
- Many overlapping sources, frequent change, strong lineage or audit needs: data vault in the integration layer
- Regulated environments requiring full historic traceability: favour vault-style history
The pragmatic combination
A common enterprise pattern is a data vault integration layer with dimensional marts published for consumption. This preserves auditability while giving analysts a model they can use, provided the marts are generated rather than hand-maintained.
Working through this in your own environment?
We help organisations make these decisions with the constraints they actually have.