GCP Professional Data Engineer Practice Question

Your retail data platform stores detailed orders data in a BigQuery dataset owned by the finance team. A marketing agency must build a Looker Studio dashboard that shows only daily revenue totals by country. The agency must never be able to read the raw order-level rows, and the finance team wants to avoid copying data or maintaining scheduled export jobs. Which approach best meets the requirements while enforcing least-privilege access?

Create a materialized view in the finance dataset that aggregates daily revenue and grant the agency BigQuery Data Viewer access on that materialized view.
Create an authorized view in a separate, shared dataset that computes daily revenue totals, grant the agency BigQuery Data Viewer on that dataset, and authorize the view to access the finance dataset.
Apply row-level security policies to the orders table and grant the agency BigQuery JobUser and Data Viewer roles on the finance dataset.
Schedule a daily query that writes revenue aggregates to a new BigQuery table in a shared dataset and grant the agency BigQuery Data Viewer access on that table.

Report Issue

Answer Description

Creating an authorized view lets you expose a query's results (for example, daily revenue totals) while withholding direct access to the underlying orders table. The view is placed in a separate dataset that is shared with the agency, and the view itself is granted read access to the source dataset. Because users query the view through the BigQuery API, they can see only the aggregated results; attempts to reference the base table are blocked. This satisfies the requirement without duplicating data or adding operational overhead.

A standard materialized view (first option) would still require the agency to have access to the base table's dataset, violating the security requirement. Writing aggregates to a new table with a scheduled query (third option) duplicates data and introduces ongoing maintenance. Row-level security or column-level masking (fourth option) cannot guarantee that ad-hoc queries won't expose raw rows, and masking is column-, not row-centric, so it does not address the need to hide individual order rows.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.