Under which conditions is denormalization appropriate in MDC3?

• A. Denormalization should always be avoided.
• B. Denormalization reduces data integrity.
• C. Denormalization is used to store more data in a single table, which is never recommended.
• D. Denormalization is used to improve read performance and is applied selectively when reporting demands, while normalization remains for integrity.

Answer: (D)

Denormalization is about trading some data integrity safeguards for faster read performance by duplicating data or combining related tables. The main reason to use it is when queries for reporting or analytics are read-heavy and joins across many tables would slow things down. By denormalizing, you reduce the need for complex joins and can retrieve the needed information more quickly.

Because redundancy is introduced, updates can become more error-prone if every place that stores duplicated data isn’t kept in sync. That’s why this technique is applied selectively: you keep normalization in place to protect data integrity where it matters for transactional operations, and you denormalize only where there’s a clear performance benefit and you have a reliable way to maintain consistency—often with controlled update paths, constraints, or tools like materialized views.

So the best description is that denormalization is used to improve read performance and is applied selectively when reporting demands exist, while normalization remains to preserve integrity.