問題文
A product dimension has 40 attributes, of which a category hierarchy of three levels changes names occasionally. The reporting workload is read-heavy and served to a BI tool. Which reasoning supports keeping the dimension denormalized as a star rather than snowflaking the hierarchy?
選択肢
- Denormalized dimensions guarantee that a category name change propagates automatically to every historical fact row that referenced the old name, which normalizing the hierarchy would prevent.
- A snowflake schema cannot represent hierarchies deeper than two levels, so the three-level hierarchy forces a star.
- Read queries stay simple with fewer joins, and the occasional name change can be handled by updating the dimension rather than by normalizing it.
- Denormalizing eliminates the need for surrogate keys, which is the main source of join cost in a star schema.