Pass Microsoft DP-800 Exam in First Attempt Guaranteed!
Get 100% Latest Exam Questions, Accurate & Verified Answers!
30 Days Free Updates, Instant Download!
Microsoft DP-800 Practice Test 37
Question 360
A supplier sends one JSON payload containing an array of shipment items and the importer currently saves the array as an opaque string. Which concept should guide the database developer in this situation?
Explanation
Correct Answer
OPENJSON supports transforming structured arrays into queryable relational data.
Incorrect Answers
MATCH expresses graph patterns across documented node and edge relationships. OPENJSON supports transforming structured arrays into queryable relational data.
CTEs improve composability and readability but are not automatically materialized performance caches. OPENJSON supports transforming structured arrays into queryable relational data.
JSON_VALUE returns scalar text semantics, so downstream typing should match the intended operation. OPENJSON supports transforming structured arrays into queryable relational data.
Question 361
An application wants new FAQ entries to receive embeddings soon after insertion without embedding generation blocking the user-facing SQL transaction. What should the database developer verify to confirm the suspected cause?
Explanation
Correct Answer
Client inserts slow down when synchronous model calls run inside the database write path. Asynchronous embedding generation decouples inference latency from transactional writes.
Incorrect Answers
An analyst cannot determine whether two vectors came from the same source revision or embedding model. Asynchronous embedding generation decouples inference latency from transactional writes. Client inserts slow down when synchronous model calls run inside the database write path.
The vector table retains deleted product entries and lacks vectors for recent updates. Asynchronous embedding generation decouples inference latency from transactional writes. Client inserts slow down when synchronous model calls run inside the database write path.
An outage at the model endpoint causes ordinary customer updates to fail or stall. Asynchronous embedding generation decouples inference latency from transactional writes. Client inserts slow down when synchronous model calls run inside the database write path.
Question 362
A multitenant database serves a tiny customer and a giant enterprise through the same stored procedure; one cached plan is poor for the other. Select TWO: the best action and the outcome that confirms it worked.
Explanation
Correct Answers
Both tenant sizes meet their respective query latency goals under realistic mixed traffic. The accompanying action is to evaluate parameter-sensitive plan features and validated mitigation for the engine version.
Evaluate parameter-sensitive plan features and validated mitigation for the engine version. Both tenant sizes meet their respective query latency goals under realistic mixed traffic.
Incorrect Answers
The analysis identifies specific statements and a tested change reduces peak resource use. Both tenant sizes meet their respective query latency goals under realistic mixed traffic. The corresponding check is whether both tenant sizes meet their respective query latency goals under realistic mixed traffic.
Evaluate row-versioning-based read committed behavior with its storage and consistency requirements. Both tenant sizes meet their respective query latency goals under realistic mixed traffic. Evaluate parameter-sensitive plan features and validated mitigation for the engine version.
Analyze the deadlock graph and standardize access order or shorten transactions based on evidence. Both tenant sizes meet their respective query latency goals under realistic mixed traffic.
Question 363
A vector table stores embeddings from one model and a replacement model outputs a different dimension count but the team plans to mix both in the same index. What is the MOST appropriate initial remediation?
Explanation
Correct Answer
Create a compatible versioned embedding migration with dimension checks and rebuild the affected index where required. Vector dimensions and model spaces must be compatible for meaningful similarity calculations.
Incorrect Answers
Move inference to an appropriate asynchronous change processing flow or tightly bound the operation after evaluating failure risk. Vector dimensions and model spaces must be compatible for meaningful similarity calculations. Create a compatible versioned embedding migration with dimension checks and rebuild the affected index where required.
Version the model configuration and compare outputs, latency, and contracts before controlled replacement. Vector dimensions and model spaces must be compatible for meaningful similarity calculations. Create a compatible versioned embedding migration with dimension checks and rebuild the affected index where required.
Use an approved secure identity or secret-management mechanism supported by the SQL platform and model endpoint. Vector dimensions and model spaces must be compatible for meaningful similarity calculations. Create a compatible versioned embedding migration with dimension checks and rebuild the affected index where required.
Question 364
A subscription report needs the difference between current and prior month revenue per product without a complex self-join. Which clue best distinguishes this issue from similar incidents?
Explanation
Correct Answer
The existing self-join misses a product because its previous month is absent. Window offset functions simplify ordered comparisons but do not create missing time periods.
Incorrect Answers
An item containing quotes causes the generated JSON string to fail a parser test. Window offset functions simplify ordered comparisons but do not create missing time periods. The existing self-join misses a product because its previous month is absent.
The report shows a price amount from a different region than the chosen maximum date. Window offset functions simplify ordered comparisons but do not create missing time periods. The existing self-join misses a product because its previous month is absent.
A basic replace misses some token variations and leaves identifiers in test messages. Window offset functions simplify ordered comparisons but do not create missing time periods. The existing self-join misses a product because its previous month is absent.
Question 365
A payment archive table grows monthly and retention jobs currently delete millions of old rows while blocking online activity. What distinction makes the recommended approach appropriate?
Explanation
Correct Answer
Partitioning can make range maintenance practical when keys and boundaries match retention policy.
Incorrect Answers
Uniqueness keys must encode the business scope of identity rather than an arbitrary column. Partitioning can make range maintenance practical when keys and boundaries match retention policy.
Referential constraints express durable relationships more reliably than application-only checks. Partitioning can make range maintenance practical when keys and boundaries match retention policy.
Physical storage decisions should follow transaction shapes rather than blanket performance claims. Partitioning can make range maintenance practical when keys and boundaries match retention policy.
Question 366
A batch embedding job calls a model endpoint without rate-limit handling and silently skips rows when throughput is exceeded. Which technical principle explains the correct response?
Explanation
Correct Answer
Reliable embedding generation needs operational controls beyond issuing model requests.
Incorrect Answers
Embedding stores need a maintenance strategy that follows authoritative source changes. Reliable embedding generation needs operational controls beyond issuing model requests.
Change capture can drive incremental embeddings if consumers handle event ordering and retention. Reliable embedding generation needs operational controls beyond issuing model requests.
Model modalities and preprocessing must match the stored evidence types. Reliable embedding generation needs operational controls beyond issuing model requests.
Question 367
A shared SQL table stores tickets for many clients and an application bug omits its WHERE tenant predicate. How should the SQL engineering team confirm the issue is truly resolved?
Explanation
Correct Answer
Cross-tenant negative tests block forbidden rows even when the application filter is omitted. The operating rule is: Row-level security enforces data visibility in the database independently of application query discipline.
Incorrect Answers
Authorized model calls succeed with least privilege and no plaintext key is exposed to unrelated readers. The operating rule is: Row-level security enforces data visibility in the database independently of application query discipline. Cross-tenant negative tests block forbidden rows even when the application filter is omitted.
Attempted destructive actions are blocked and permitted read tools function for authorized identities. The operating rule is: Row-level security enforces data visibility in the database independently of application query discipline. Cross-tenant negative tests block forbidden rows even when the application filter is omitted.
The application connects with its scoped identity while no reusable database password is stored in code. The operating rule is: Row-level security enforces data visibility in the database independently of application query discipline. Cross-tenant negative tests block forbidden rows even when the application filter is omitted.
Question 368
A release changes a table consumed by an older mobile API but no backward-compatibility window or rollback plan exists. Which outcome shows the requirement is met without introducing a new risk?
Explanation
Correct Answer
Both old and new clients function through the transition and rollback works in a rehearsed test.
Incorrect Answers
Deployment succeeds only for its intended environment while cross-environment tests fail safely. Both old and new clients function through the transition and rollback works in a rehearsed test.
Deployment preserves required values or uses a documented approved transformation without silent truncation. Both old and new clients function through the transition and rollback works in a rehearsed test.
The pipeline rejects the faulty change and accepts the corrected procedure with verified results. Both old and new clients function through the transition and rollback works in a rehearsed test.
Question 369
A finance platform wants SQL records from supplier invoices, but its chosen general chat model outputs inconsistent field names and free-form commentary. Which observation is the strongest evidence for the reported problem?
Explanation
Correct Answer
The same document produces different field names and invalid JSON across retries. Model choice should match structured output requirements rather than general language fluency alone.
Incorrect Answers
An outage at the model endpoint causes ordinary customer updates to fail or stall. Model choice should match structured output requirements rather than general language fluency alone. The same document produces different field names and invalid JSON across retries.
New vector inserts fail dimension validation or comparisons return inconsistent similarity results. Model choice should match structured output requirements rather than general language fluency alone. The same document produces different field names and invalid JSON across retries.
The vector table retains deleted product entries and lacks vectors for recent updates. Model choice should match structured output requirements rather than general language fluency alone. The same document produces different field names and invalid JSON across retries.
Question 370
An API accepts arbitrary malformed JSON in a text field and downstream reporting intermittently fails during JSON parsing. Which observation is the strongest evidence for the reported problem?
Explanation
Correct Answer
Test inserts include invalid JSON syntax that is accepted without validation and later breaks consumers. Data shape and validity should be enforced as close to ingestion as practical.
Incorrect Answers
Investigators cannot determine what preference was effective before the latest update from the current table. Data shape and validity should be enforced as close to ingestion as practical. Test inserts include invalid JSON syntax that is accepted without validation and later breaks consumers.
A direct SQL write bypasses the front-end validator and saves invalid guest counts. Data shape and validity should be enforced as close to ingestion as practical. Test inserts include invalid JSON syntax that is accepted without validation and later breaks consumers.
The current query extracts a property for every row even when a small subset matches. Data shape and validity should be enforced as close to ingestion as practical. Test inserts include invalid JSON syntax that is accepted without validation and later breaks consumers.
Developing AI-Enabled Database Solutions (DP-800) Practice Tests
- Microsoft DP-800 Practice Test 35 - 10 Questions (Last Reviewed 13-Oct-2026)
- Microsoft DP-800 Practice Test 36 - 10 Questions (Last Reviewed 12-Oct-2026)
- Microsoft DP-800 Practice Test 37 - 10 Questions (Last Reviewed 11-Oct-2026)
- Microsoft DP-800 Practice Test 38 - 10 Questions (Last Reviewed 10-Oct-2026)
- Microsoft DP-800 Practice Test 39 - 10 Questions (Last Reviewed 09-Oct-2026)