Unity Catalog: One Governance Layer Across Every Databricks Workspace

Most Databricks estates start with one workspace and end up with five: one per business unit, one per cloud region, one somebody spun up for a proof of concept and never tore down. Each workspace used to carry its own permissions, its own copy of shared reference tables, and its own audit trail nobody else could see.

Unity Catalog sits underneath every workspace as one governance layer. Tables, views, volumes, functions, and models register under a three-level namespace, catalog.schema.object, and that namespace is shared across every workspace attached to the same metastore. A permission granted once applies everywhere the object is used, not once per workspace.

That's what actually fixes the cost problem. Teams stop maintaining duplicate copies of the same data because "our workspace can't see their catalog." A table three teams need gets accessed, not copied three times. Lineage and audit logs are unified too, so tracing which job or dashboard is reading a given table doesn't mean checking five separate workspace logs by hand.

Unity Catalog also supports the OpenSharing protocol, which lets teams share live data and AI assets with other organizations or across clouds without exporting or copying it first.

One practical note: Unity Catalog has been on by default for any workspace created since November 2023. If any of your workspaces predate that, they're worth checking first, since an ungoverned workspace is usually where duplicate storage and untracked spend hide.