Unity Catalog: governance from day one

Governance has an image problem: it sounds like the brake, not the accelerator. On a Lakehouse the opposite is true. Without a central governance layer, your platform grows into a sprawl of disconnected workspaces where nobody knows who can access what, or where data came from. Unity Catalog fixes that structurally.

One catalog across all workspaces

Unity Catalog lifts governance out of the individual workspace to the level of your whole account. You manage grants, lineage and quality in one place, with a clear hierarchy: catalog → schema → table. A sensible layout, for example a catalog per domain or per environment, saves you from restructuring later.

Access that stays explainable

Grant permissions to groups, not individuals, and use fine-grained controls where you need them: row filters and column masks keep sensitive data protected without separate copies. The result is access you can explain to an auditor, and that scales with your organisation.

Lineage and audit, for free

Because every query goes through Unity Catalog, Databricks captures lineage automatically, from source file to dashboard and down to column level. That is invaluable for impact analysis (“what does this change touch?”) and for tracing the source of a wrong number. Audit logs additionally tell you who accessed what, and when.

Capture it as code

Don’t click governance together in the UI. With the Databricks Terraform provider you manage catalogs, schemas, grants and external locations as Infrastructure as Code. That makes your setup reproducible, reviewable through pull requests, and identical across dev, test and production. Why Terraform and not Bicep? I’ve laid out that choice separately.

Start early, start small

The biggest mistake is saving governance “for later”. Set up Unity Catalog as soon as the first tables land: it costs little then, whereas restructuring afterwards is expensive and painful. Starting small is fine, but start.