DAA-C01 Question 212
Single answerCollect related dataA retail analytics team stores customer purchases in a SALES_FACT table and customer support interactions in a SUPPORT_CASES table. Analysts frequently need to correlate purchases with support cases for the same customer and then explore related product and campaign details from PRODUCT_DIM and CAMPAIGN_DIM. The team wants a reusable semantic layer in Snowsight that helps analysts collect related data interactively without rewriting joins each time, while preserving the underlying table relationships for consistent analysis. Which approach best meets this requirement?
- A
Create a view that fully joins all four tables and have analysts query the view directly for every use case.
- B
Create a semantic model that defines relationships among the fact and dimension tables so analysts can navigate related data consistently in Snowsight.
- C
Create a materialized view on SALES_FACT and use search optimization on SUPPORT_CASES so analysts can manually join the objects when needed.
- D
Create a dynamic table that denormalizes SALES_FACT, SUPPORT_CASES, PRODUCT_DIM, and CAMPAIGN_DIM into a single reporting table refreshed on a schedule.
Show answer and explanation
Correct answer: B
Explanation
The key requirement is a reusable semantic layer in Snowsight that lets analysts collect related data across multiple related tables without repeatedly hand-coding joins. The best answer is to create a semantic model with defined relationships among the fact and dimension tables. In Snowflake, semantic modeling is intended to capture business logic and relationships so users can analyze related data consistently and interactively. By contrast, views, materialized views, and dynamic tables can simplify or optimize data access, but they do not provide the same governed semantic abstraction for relationship-driven exploration. This aligns with Snowflake best practices of separating semantic/business logic from physical storage and using semantic models to make related data easier for analysts to discover and query consistently.
- A. Incorrect.
This is not the best choice for the stated requirement. A single wide view can simplify some queries, but it hard-codes one join pattern and can become difficult to maintain as analytical needs evolve. It also does not provide a reusable semantic layer for interactively collecting related data in a governed way; it simply exposes a prejoined dataset.
- B. Correct.
This is correct. A semantic model is designed to represent business entities, measures, dimensions, and relationships so analysts can explore related data consistently without repeatedly rewriting joins. For a scenario focused on collecting related data across facts and dimensions in Snowsight, defining relationships in a semantic model best supports governed, reusable, interactive analysis.
- C. Incorrect.
This is incorrect because these features address performance and storage patterns, not semantic navigation of related data. A materialized view can improve query performance for specific patterns, and search optimization can accelerate selective lookups, but analysts would still need to understand and write joins manually. That does not satisfy the requirement for a reusable semantic layer.
- D. Incorrect.
This is a plausible approach for some reporting workloads, but it is not the best fit here. A dynamic table can maintain a transformed dataset, but denormalizing everything into one table reduces flexibility, can duplicate data, and does not preserve table relationships as a semantic construct for interactive exploration. It solves a data preparation problem more than a governed semantic modeling problem.