Google Professional Data Engineer Question 145
Single answerGoogle Cloud PlatformYou are designing a Google Cloud-based data warehouse for an e-commerce platform that processes millions of transactions daily. The data will be consumed by analytics teams for reporting and predictive modeling. You need to decide the degree of normalization for storing transaction data. Which factor should most influence your decision?
- A
The need for minimizing data redundancy to reduce storage costs
- B
Optimizing query performance for large-scale analytical queries
- C
Ensuring the database schema adheres strictly to third normal form (3NF)
- D
The ability to scale horizontally with minimal schema changes
Show answer and explanation
Correct answer: B
Explanation
When designing a data warehouse for analytics, query performance is the most significant consideration. Denormalization is often used in analytical workloads to reduce the complexity of joins and improve query performance. While normalization is beneficial for transactional systems, it can negatively impact performance in analytical systems where large-scale queries are executed frequently.
- A. Incorrect.
While minimizing data redundancy can be a benefit of normalization, storage costs are often less significant in modern cloud environments, especially with compression technologies. Therefore, this is not the primary factor in deciding normalization for analytics.
- B. Correct.
Optimizing query performance is the most critical factor in determining the degree of normalization for large-scale analytics. Denormalized schemas, such as star or snowflake schemas, are often preferred for analytical workloads to improve query performance.
- C. Incorrect.
Adhering strictly to 3NF is more relevant for transactional workloads (OLTP) where data integrity and update efficiency are critical. For analytics (OLAP), denormalization is often preferred.
- D. Incorrect.
While scalability is important, it is typically addressed through other architectural decisions, such as sharding or using BigQuery's serverless capabilities, rather than the degree of normalization.