Databricks Data Engineer Professional study materials
Databricks Data Engineer Professional study resources

Databricks Data Engineer Professional: Data Governance

Databricks Data Engineer Professional · Work through your exam, one topic at a time.

All guide sections

Data Governance

The object hierarchy

The Unity Catalog hierarchy - metastore → catalog → schema → table/view/volume/function - is the backbone of governance questions. Know that managed tables live in the managed storage root of the metastore and are dropped with their data, while external tables point at an external location and survive a DROP TABLE as files. Managed volumes versus external volumes follow the same logic.

Metadata and discoverability

Comments, tags, and properties are queryable through information_schema and system.information_schema; column tags drive classification and (with ABAC) policy targeting. Lineage is captured automatically for UC objects and exposed in the UI and system.access.table_lineage; column-level lineage is available for many query types. Understand that lineage for non-UC paths (DBFS mounts, legacy Hive metastore) is limited - a common exam theme about why lineage is missing.

Ownership, transfer, and lifecycle

Every object has an owner; a metastore admin can transfer ownership. Deleting a catalog is not the same as deleting a schema, and CASCADE matters. External locations require a storage credential that wraps a cloud IAM role, and any external table path must be covered by an external location - otherwise creation fails.

Rule of thumb: govern at the catalog/schema level with tags and ownership, secure with external locations, and never use DBFS mounts for new governed data.

Governance also covers data sharing contracts (providers, shares, recipients) and clean-room style collaboration, but always through UC objects.