You are designing a RAG solution that must handle multiple concurrent requests from users. Each request involves retrieving data, building a prompt, and calling the language model. You are concerned about performance and rate limits. What is the best approach?

Use a queue to process requests asynchronously and batch calls to the language model where possible.

Call the language model synchronously for each request without any throttling.

Use a linked server to handle concurrency.

Use a single large prompt for all concurrent requests.

A legal firm stores contracts in a SQL Server database. They want to build a system where lawyers can ask natural-language questions about contract clauses, and the system returns answers that cite the specific contract text. The firm requires that the answers be grounded in the stored contract data, not generated from the model's general knowledge. Which approach should you recommend?

Use a model that has been pre-trained on legal documents and rely on its general knowledge.

Use a fine-tuned model that has been trained on the contract database.

Use a RAG pattern: retrieve relevant contract clauses from the database, include them in the prompt, and send the prompt to the model.

Use a model with a large context window and send the entire contract database in the prompt.

After calling sp_invoke_external_rest_endpoint, the response is stored in a variable of type NVARCHAR(MAX). The response contains a JSON object with a property named 'choices' that is an array. Each element has a 'text' property. You need to extract the first text value. What should you do?

Use JSON_QUERY with the path '$.choices[0].text'.

Use OPENJSON with the path '$.choices' and then select the first row.

Use the SUBSTRING function to manually parse the response.

Use JSON_VALUE with the path '$.choices[0].text'.

Microsoft Certified:SQL AI Developer Associate

Domain 3Objective 3

Design and Implement Retrieval-Augmented Generation (RAG) DP-800 Practice Questions (Page 7)

Part of the Implement AI capabilities in database solutions domain, which accounts for 25–30% of the DP-800 exam. Microsoft does not publish an official question count, but from its 120-minute exam (~50–80 total, ~13–24 in this domain), expect 4–8 from this objective — we provide 33 practice questions to prepare you well beyond it. (estimate)

33questions here

7free pages

5concepts

25–30%of the exam

Questions 31–33

31expert · hard
You are designing a RAG solution that must handle multiple concurrent requests from users. Each request involves retrieving data, building a prompt, and calling the language model. You are concerned about performance and rate limits. What is the best approach?
Select an answer first
32application · medium
A legal firm stores contracts in a SQL Server database. They want to build a system where lawyers can ask natural-language questions about contract clauses, and the system returns answers that cite the specific contract text. The firm requires that the answers be grounded in the stored contract data, not generated from the model's general knowledge. Which approach should you recommend?
Select an answer first
33application · medium
After calling sp_invoke_external_rest_endpoint, the response is stored in a variable of type NVARCHAR(MAX). The response contains a JSON object with a property named 'choices' that is an array. Each element has a 'text' property. You need to extract the first text value. What should you do?
Select an answer first

Finished these 3 questions?

Review the revealed explanations, or continue through the curriculum.

No more pagesBack to DP-800

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Microsoft. “DP-800” is a trademark of its owner, used for identification only.

Design and Implement Retrieval-Augmented Generation (RAG) DP-800 Practice Questions (Page 7)

Questions 31–33

You are designing a RAG solution that must handle multiple concurrent requests from users. Each request involves retrieving data, building a prompt, and calling the language model. You are concerned about performance and rate limits. What is the best approach?

After calling sp_invoke_external_rest_endpoint, the response is stored in a variable of type NVARCHAR(MAX). The response contains a JSON object with a property named 'choices' that is an array. Each element has a 'text' property. You need to extract the first text value. What should you do?

Keep studying — Implement AI capabilities in database solutions