Databricks Machine Learning Associate Question 425
Select 3You are leading a team of data scientists managing multiple Databricks workspaces within your organization's Unity Catalog-enabled account. You are tasked with creating a feature store for a machine learning project that will be used across different workspaces. Why might you choose to create a feature store table at the account level in Unity Catalog rather than at the workspace level?
- A
Feature store tables at the account level enable centralized access control and governance across multiple workspaces.
- B
Creating feature store tables at the account level helps reduce storage costs by avoiding data duplication across workspaces.
- C
Account-level feature store tables ensure that features are always automatically versioned across all workspaces.
- D
Using account-level feature store tables allows teams to collaborate across workspaces without managing workspace-specific dependencies.
- E
Feature store tables at the account level automatically enforce higher data processing speeds compared to workspace-level tables.
Show answer and explanation
Correct answers: A, B, D
Explanation
Creating feature store tables at the account level in Unity Catalog offers several advantages, particularly for organizations that manage multiple workspaces. Centralized access control ensures consistent governance, while avoiding data duplication cuts down on storage costs. Additionally, account-level tables facilitate collaboration across workspaces by removing the need for workspace-specific dependencies. However, certain features like versioning and data processing speeds are not directly tied to the level at which the feature store is created.
- A. Correct.
Centralized access control and governance is a key benefit of account-level tables. Unity Catalog allows you to define permissions at the account level, ensuring consistent policies across workspaces.
- B. Correct.
Avoiding data duplication helps reduce storage costs because features stored at the account level can be reused across multiple workspaces without creating multiple copies.
- C. Incorrect.
While versioning is a crucial feature of the Databricks Feature Store, it is not inherently tied to the account-level or workspace-level scope. Versioning depends on how features are managed, not where they are stored.
- D. Correct.
Collaboration across workspaces is easier with account-level tables because teams do not need to replicate or manage workspace-specific configurations to share features.
- E. Incorrect.
Data processing speeds are determined by factors like infrastructure and computation resources, not by whether the feature store tables are at the account or workspace level.