Google Professional Data Engineer Question 169
Select 3Google Cloud PlatformYour organization is implementing a data mesh architecture on Google Cloud to empower different business units to manage their own datasets while ensuring proper governance and discoverability. As the Data Engineer, you need to set up the foundational components to enable this architecture. Which combination of Google Cloud tools should you use to meet these requirements?
- A
Use Dataplex to create logical data domains and enforce governance policies across datasets.
- B
Use Data Catalog to enable metadata management and dataset discoverability across the organization.
- C
Use BigQuery to replicate all datasets into a central data warehouse for unified access.
- D
Use Cloud Storage as the primary storage for raw data and to decentralize data ownership.
- E
Use Dataproc for building a centralized data pipeline to process all datasets.
Show answer and explanation
Correct answers: A, B, D
Explanation
To build a data mesh on Google Cloud, you need tools that support decentralization, governance, and discoverability. Dataplex enables logical data domains and governance, Data Catalog supports metadata management and discoverability, and Cloud Storage provides a decentralized storage layer for raw data. BigQuery and Dataproc, while powerful, are not suitable for the decentralized and domain-oriented nature of a data mesh in this scenario.
- A. Correct.
Correct. Dataplex is specifically designed for organizing data into logical domains, managing governance policies, and enabling a data mesh architecture.
- B. Correct.
Correct. Data Catalog provides metadata management and discoverability features, which are critical for the self-service aspects of a data mesh.
- C. Incorrect.
Incorrect. Replicating all datasets into a central warehouse would contradict the decentralized principles of a data mesh.
- D. Correct.
Correct. Cloud Storage can be used as a decentralized storage layer to enable business units to own and manage their raw datasets.
- E. Incorrect.
Incorrect. While Dataproc is useful for processing data, building a centralized pipeline does not align with the distributed nature of a data mesh.