Databricks Data Engineer Associate Question 531
Select 2A company is using Databricks Unity Catalog to manage its data lake. The organization consists of multiple business units such as Marketing, Sales, and Finance, each requiring secure access to their respective datasets. What is the best practice for segregating access across these business units?
- A
Create a separate catalog for each business unit, such as 'marketing_catalog', 'sales_catalog', and 'finance_catalog', to isolate their data.
- B
Use the same catalog for all business units but segregate data using schema-level permissions.
- C
Assign row-level access controls within a single schema to segregate data for each business unit.
- D
Use a separate catalog for each business unit and implement schema-level permissions within those catalogs to further restrict access.
Show answer and explanation
Correct answers: A, D
Explanation
The best practice for segregating business units in Databricks Unity Catalog is to create separate catalogs for each unit, ensuring clear isolation of data. Additionally, schema-level permissions within those catalogs can be implemented to provide more granular control over access. This approach simplifies access management while maintaining strict segregation across units.
- A. Correct.
Correct. Creating separate catalogs for each business unit is a best practice to ensure data segregation at the highest level, as catalogs are the top-level container in Unity Catalog.
- B. Incorrect.
Incorrect. Using the same catalog for all business units can lead to a more complex permissions model and increases the risk of accidental access across business units.
- C. Incorrect.
Incorrect. Row-level access controls are more appropriate for fine-grained access within a single dataset, not for segregating data across entire business units.
- D. Correct.
Correct. Using a separate catalog for each business unit ensures high-level segregation, and applying schema-level permissions within those catalogs provides additional control over access to specific datasets.