Databricks Data Engineer Associate · 15% practice allocation

Databricks Data Engineer Associate: Governance and Security Practice Questions

5 free sample questions from the Governance and Security domain, with answers and explanations. The full Data Engineer Associate bank has 300 questions across 7 domains. Questions are scenario-based and written from the official Databricks exam guide.

1.

A company stores raw CSV files in an AWS S3 bucket that is managed by their data platform team. A data engineer wants to register this S3 location as a table in Unity Catalog so that if the table is dropped, the CSV files are NOT deleted. Which table type should the engineer create?

  • A. An external table pointing to the S3 path✓ Correct
  • B. A managed table pointing to the S3 path
  • C. A materialized view over the S3 path
  • D. A Delta Live Tables dataset backed by the S3 path
Explanation

An external table in Unity Catalog is backed by a user-specified storage location (like an S3 path). When an external table is dropped, only the metadata is removed from the metastore; the underlying data files at the external location are left intact. This is exactly the behavior required. A managed table would cause data files to be deleted when the table is dropped, which violates the requirement. A Delta Live Tables dataset is a pipeline construct and is not the same as directly registering a pre-existing CSV location as a table. A materialized view is a computed, read-only object derived from a query — it does not directly map raw CSV files from an external path as a regular queryable table.

2.

A data engineering team wants to organize their Databricks environment so that each business unit (Finance, Marketing, HR) has its own isolated namespace for tables and views, while all units share a single Unity Catalog metastore. Which Unity Catalog object should they create one of per business unit to achieve this isolation?

  • A. Metastore
  • B. Catalog✓ Correct
  • C. Table
  • D. Schema
Explanation

A Catalog is the correct level of isolation in the Unity Catalog three-level namespace (catalog.schema.table). One catalog per business unit provides a clean namespace boundary while all catalogs still reside under the single account-level metastore. A Metastore is wrong because only one metastore exists per Databricks account region and cannot be multiplied per business unit. A Schema is too granular — it lives inside a catalog and is used to group related tables within a business unit, not to isolate units from each other. A Table is a data object, not a namespace container.

3.

A table must expose all rows to admins but only a region's rows to regional analysts, without duplicating the table. Which Unity Catalog feature supports this?

  • A. A row filter (row-level security)✓ Correct
  • B. VACUUM
  • C. OPTIMIZE ZORDER
  • D. A separate physical copy per region
Explanation

A row filter applies row-level security so different principals see different rows of the same table. VACUUM/OPTIMIZE (B, C) are maintenance, and physical copies (D) defeat the single-source goal.

4.

A Unity Catalog managed table is dropped using DROP TABLE. What happens to the underlying data files?

  • A. The metadata is removed from the metastore, but the data files remain in the managed storage location indefinitely.
  • B. Both the metadata and the underlying data files are permanently deleted by Unity Catalog.✓ Correct
  • C. The data files are converted to an external table automatically so data is preserved.
  • D. The data files are moved to a recycle bin location in the metastore's root storage for 30 days before deletion.
Explanation

For a Unity Catalog managed table, Databricks manages the full lifecycle — when the table is dropped, Unity Catalog deletes both the metadata entry in the metastore AND the physical data files in the managed storage location. This is the key distinction from external tables. Option A is incorrect because it describes the behavior of an external table, not a managed one. Option D is a misconception — there is no automatic 30-day recycle bin for managed table data files (though Delta time travel can retain older versions for a configured retention window, the files are not moved to a separate recycle bin). Option C is incorrect; dropping a managed table does not auto-convert it to an external table.

5.

A data engineer runs REVOKE SELECT ON TABLE catalog1.schema1.orders FROM reporting_group. A member of reporting_group then tries to query the table. Which of the following statements accurately describes what happens next? (Select TWO)

  • A. The query fails because the reporting_group no longer holds the SELECT privilege on the table.✓ Correct
  • B. If the user is also a direct member of another group that still has SELECT on the table, the query succeeds.✓ Correct
  • C. REVOKE automatically cascades and removes SELECT from all schemas in catalog1.
  • D. The user's existing Delta cache of the table data allows the query to succeed despite the revoked privilege.
  • E. The query succeeds because REVOKE only removes future privilege grants, not existing session access.
Explanation

Option A is correct: REVOKE immediately removes the SELECT privilege from reporting_group, so the next query attempt by a member of that group (acting through that group's privilege) will fail with an authorization error. Option B is also correct: Unity Catalog evaluates privileges using union of all group memberships — if the user belongs to another group that still has SELECT on the table, access is granted through that group's privilege. Option E is incorrect: REVOKE takes effect immediately for subsequent queries; there is no concept of 'grandfathering' existing session access. Option C is incorrect: REVOKE is a targeted operation on the specified securable object and does not cascade to other objects in the catalog or schema. Option D is incorrect: The Delta cache stores query result data for performance, but Unity Catalog privilege checks happen at query execution time on the engine regardless of cache state; a user without SELECT cannot query the table.

Test yourself across every Data Engineer Associate domain

The free Data Engineer Associate practice test mixes about 20 questions weighted like the exam. The full 300-question bank adds instant grading, flashcards and a timed mock exam.

Study this domain

Governance and Security study guide (Data Engineer Associate) →

Other Data Engineer Associate domains

← All Databricks Data Engineer Associate study resources