Microsoft DP-800 Practice Test 27 - Prepaway
Practice Exams:

Pass Microsoft DP-800 Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers!

30 Days Free Updates, Instant Download!

Verified By Experts
Premium File
$69.99 $76.99

DP-800 Premium ETE File


  • Premium File 101 Questions & Answers. Last Update: Nov 06, 2026
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft DP-800 Practice Test 27

Question 260

A SQL database invokes an external model that is replaced by the provider without a tested compatibility plan. How should the SQL engineering team confirm the issue is truly resolved?

  1. A staged migration passes regression benchmarks and retains a clear rollback route.
  2. Authorized calls work and no unapproved identity can use the integration to retrieve or generate protected content.
  3. The index can be audited and stale or incompatible embeddings are identified and replaced.
  4. A replay-safe feed propagates additions, updates, and deletions without a full rebuild.

Explanation

Correct Answer

A:

A staged migration passes regression benchmarks and retains a clear rollback route. A SQL database invokes an external model that is replaced by the provider without a tested compatibility plan.

Incorrect Answers

B:

Authorized calls work and no unapproved identity can use the integration to retrieve or generate protected content. A SQL database invokes an external model that is replaced by the provider without a tested compatibility plan. A staged migration passes regression benchmarks and retains a clear rollback route.

C:

The index can be audited and stale or incompatible embeddings are identified and replaced. A SQL database invokes an external model that is replaced by the provider without a tested compatibility plan. A staged migration passes regression benchmarks and retains a clear rollback route.

D:

A replay-safe feed propagates additions, updates, and deletions without a full rebuild. A SQL database invokes an external model that is replaced by the provider without a tested compatibility plan. A staged migration passes regression benchmarks and retains a clear rollback route.

Question 261

A developer asks Copilot to fix a production query and reports that the database is repaired because the chat returned corrected code. Which corrective action best fits the observed symptoms?

  1. Replace live values with safe synthetic examples and enforce repository secret and privacy scanning.
  2. Scope the connector to approved lakehouse data and verify the selected identity and workspace permissions.
  3. Separate assistant suggestions from approved, executed changes and verify actual database state through controlled deployment.
  4. Create reviewed repository instruction files that specify migration, test, and secure coding conventions without embedding secrets.

Explanation

Correct Answer

C:

Separate assistant suggestions from approved, executed changes and verify actual database state through controlled deployment. Generated SQL text is not evidence of a successful database change.

Incorrect Answers

A:

Replace live values with safe synthetic examples and enforce repository secret and privacy scanning. Generated SQL text is not evidence of a successful database change. Separate assistant suggestions from approved, executed changes and verify actual database state through controlled deployment.

B:

Scope the connector to approved lakehouse data and verify the selected identity and workspace permissions. Generated SQL text is not evidence of a successful database change. Separate assistant suggestions from approved, executed changes and verify actual database state through controlled deployment.

D:

Create reviewed repository instruction files that specify migration, test, and secure coding conventions without embedding secrets. Generated SQL text is not evidence of a successful database change. Separate assistant suggestions from approved, executed changes and verify actual database state through controlled deployment.

Question 262

A SQL MCP server advertises a generic execute-query tool that can run destructive statements using elevated credentials. Select TWO: the best action and the outcome that confirms it worked.

  1. The endpoint returns permitted entities and rejects unauthorized resource and operation combinations.
  2. Configure appropriate SQL auditing to a protected destination with tested retention and access reviews.
  3. Attempted destructive actions are blocked and permitted read tools function for authorized identities.
  4. Plan coordinated key rotation and recovery testing before replacing or retiring key material.
  5. Apply service authentication, tool allowlists, scoped SQL permissions, and approval for state changes.

Explanation

Correct Answers

C:

Attempted destructive actions are blocked and permitted read tools function for authorized identities. The accompanying action is to apply service authentication, tool allowlists, scoped SQL permissions, and approval for state changes.

E:

Apply service authentication, tool allowlists, scoped SQL permissions, and approval for state changes. Attempted destructive actions are blocked and permitted read tools function for authorized identities.

Incorrect Answers

A:

The endpoint returns permitted entities and rejects unauthorized resource and operation combinations. Attempted destructive actions are blocked and permitted read tools function for authorized identities.

B:

Configure appropriate SQL auditing to a protected destination with tested retention and access reviews. Attempted destructive actions are blocked and permitted read tools function for authorized identities. Apply service authentication, tool allowlists, scoped SQL permissions, and approval for state changes.

D:

Plan coordinated key rotation and recovery testing before replacing or retiring key material. Attempted destructive actions are blocked and permitted read tools function for authorized identities. Apply service authentication, tool allowlists, scoped SQL permissions, and approval for state changes.

Question 263

A hybrid search service combines full-text and vector result rankings by adding raw scores that have incompatible scales. Which observation is the strongest evidence for the reported problem?

  1. The same semantic candidate changes rank mainly because of vector magnitude rather than meaning.
  2. Global top-k candidates belong to other tenants and the postfilter removes all five.
  3. A discontinued SKU scores high on semantic relevance and appears as an orderable suggestion.
  4. One retrieval modality dominates because its arbitrary numeric score range is larger.

Explanation

Correct Answer

D:

One retrieval modality dominates because its arbitrary numeric score range is larger. RRF combines rank positions without assuming raw search score scales are comparable.

Incorrect Answers

A:

The same semantic candidate changes rank mainly because of vector magnitude rather than meaning. RRF combines rank positions without assuming raw search score scales are comparable. One retrieval modality dominates because its arbitrary numeric score range is larger.

B:

Global top-k candidates belong to other tenants and the postfilter removes all five. RRF combines rank positions without assuming raw search score scales are comparable. One retrieval modality dominates because its arbitrary numeric score range is larger.

C:

A discontinued SKU scores high on semantic relevance and appears as an orderable suggestion. RRF combines rank positions without assuming raw search score scales are comparable. One retrieval modality dominates because its arbitrary numeric score range is larger.

Question 264

An administrator uses a DML trigger to log unauthorized table changes and assumes the trigger alone prevents data exfiltration. Which observation is the strongest evidence for the reported problem?

  1. A privileged test account can still select restricted rows despite the trigger firing on writes.
  2. The second deployment creates more lookup codes than the first although the business catalog is unchanged.
  3. The plan and CPU profile show repeated row-by-row invocation dominating execution time.
  4. A plan reused from a selective parameter performs extensive key lookups for the large region.

Explanation

Correct Answer

A:

A privileged test account can still select restricted rows despite the trigger firing on writes. Triggers can react to data modifications but are not a substitute for least-privilege reads.

Incorrect Answers

B:

The second deployment creates more lookup codes than the first although the business catalog is unchanged. Triggers can react to data modifications but are not a substitute for least-privilege reads. A privileged test account can still select restricted rows despite the trigger firing on writes.

C:

The plan and CPU profile show repeated row-by-row invocation dominating execution time. Triggers can react to data modifications but are not a substitute for least-privilege reads. A privileged test account can still select restricted rows despite the trigger firing on writes.

D:

A plan reused from a selective parameter performs extensive key lookups for the large region. Triggers can react to data modifications but are not a substitute for least-privilege reads. A privileged test account can still select restricted rows despite the trigger firing on writes.

Question 265

A SQL project tracks application tables but lookup status codes are manually edited in production, causing different values across environments. Which observation is the strongest evidence for the reported problem?

  1. Clean-build logs show an unresolved referenced object while the local workstation builds.
  2. A staged test user can query sensitive salary columns without being assigned an appropriate role.
  3. The same status ID means different things in test and production after manual changes.
  4. Unit tests pass while a real GraphQL query fails on a missing relationship field.

Explanation

Correct Answer

C:

The same status ID means different things in test and production after manual changes. Static reference data can be part of database release reproducibility and should be governed.

Incorrect Answers

A:

Clean-build logs show an unresolved referenced object while the local workstation builds. Static reference data can be part of database release reproducibility and should be governed. The same status ID means different things in test and production after manual changes.

B:

A staged test user can query sensitive salary columns without being assigned an appropriate role. Static reference data can be part of database release reproducibility and should be governed. The same status ID means different things in test and production after manual changes.

D:

Unit tests pass while a real GraphQL query fails on a missing relationship field. Static reference data can be part of database release reproducibility and should be governed. The same status ID means different things in test and production after manual changes.

Question 266

Readers wait behind frequent updates in an order system and staff propose dirty reads as a quick fix for business dashboards. Select TWO: the best action and the outcome that confirms it worked.

  1. Measure actual fragmentation, page density, workload plans, and maintenance impact before a rebuild policy.
  2. Use live DMVs, wait information, and deadlock evidence to distinguish blocking chains from actual cyclic deadlocks.
  3. The new configuration reduces read blocking while reports avoid uncommitted data and temp/version costs are monitored.
  4. Evaluate row-versioning-based read committed behavior with its storage and consistency requirements.
  5. A selective maintenance plan avoids unnecessary work and preserves measured query service objectives.

Explanation

Correct Answers

C:

The new configuration reduces read blocking while reports avoid uncommitted data and temp/version costs are monitored.

D:

Evaluate row-versioning-based read committed behavior with its storage and consistency requirements. The new configuration reduces read blocking while reports avoid uncommitted data and temp/version costs are monitored.

Incorrect Answers

A:

Measure actual fragmentation, page density, workload plans, and maintenance impact before a rebuild policy. The new configuration reduces read blocking while reports avoid uncommitted data and temp/version costs are monitored.

B:

Use live DMVs, wait information, and deadlock evidence to distinguish blocking chains from actual cyclic deadlocks. The new configuration reduces read blocking while reports avoid uncommitted data and temp/version costs are monitored. Evaluate row-versioning-based read committed behavior with its storage and consistency requirements.

E:

A selective maintenance plan avoids unnecessary work and preserves measured query service goal. The new configuration reduces read blocking while reports avoid uncommitted data and temp/version costs are monitored.

Question 267

A database process sends prompts to an external model with a shared static API credential stored in plain configuration. Which next step most directly resolves this SQL solution issue?

  1. Evaluate multilingual embedding models against labeled cross-language retrieval tests and relevant privacy constraints.
  2. Embed only approved semantically relevant fields and retain a secure linkage to source permissions.
  3. Evaluate multimodal models or appropriately separated image/text embeddings with clear data handling rules.
  4. Use an approved secure identity or secret-management mechanism supported by the SQL platform and model endpoint.

Explanation

Correct Answer

D:

Use an approved secure identity or secret-management mechanism supported by the SQL platform and model endpoint. Model integration security must cover both database identity and downstream inference access.

Incorrect Answers

A:

Evaluate multilingual embedding models against labeled cross-language retrieval tests and relevant privacy constraints. Model integration security must cover both database identity and downstream inference access. Use an approved secure identity or secret-management mechanism supported by the SQL platform and model endpoint.

B:

Embed only approved semantically relevant fields and retain a secure linkage to source permissions. Model integration security must cover both database identity and downstream inference access. Use an approved secure identity or secret-management mechanism supported by the SQL platform and model endpoint.

C:

Evaluate multimodal models or appropriately separated image/text embeddings with clear data handling rules. Model integration security must cover both database identity and downstream inference access. Use an approved secure identity or secret-management mechanism supported by the SQL platform and model endpoint.

Question 268

A support API requests top five nearest results but applies an incompatible filter after the query and often receives zero results for small tenants. Which outcome shows the requirement is met without introducing a new risk?

  1. A benchmark reaches minimum relevant top-k hit rates while respecting service latency.
  2. Results remain relevant and exclude ineligible inventory under tested status changes.
  3. A pre-insert check identifies incompatible vectors and routes them to repair before indexing.
  4. Tenant-specific queries return relevant permitted results within a bounded cost.

Explanation

Correct Answer

D:

Tenant-specific queries return relevant permitted results within a bounded cost.

Incorrect Answers

A:

A benchmark reaches minimum relevant top-k hit rates while respecting service latency. Tenant-specific queries return relevant permitted results within a bounded cost.

B:

Results remain relevant and exclude ineligible inventory under tested status changes. Tenant-specific queries return relevant permitted results within a bounded cost.

C:

A pre-insert check identifies incompatible vectors and routes them to repair before indexing. Tenant-specific queries return relevant permitted results within a bounded cost.

Question 269

A SQL database invokes an external model that is replaced by the provider without a tested compatibility plan. What should the database developer verify to confirm the suspected cause?

  1. Job logs show throttled responses while the completion report claims all records were embedded.
  2. The same document produces different field names and invalid JSON across retries.
  3. The secret appears in developer-readable config and can invoke the endpoint outside the approved process.
  4. After endpoint upgrade structured responses no longer satisfy the application's JSON schema.

Explanation

Correct Answer

D:

After endpoint upgrade structured responses no longer satisfy the application's JSON schema. External models are dependencies that need version governance and compatibility testing.

Incorrect Answers

A:

Job logs show throttled responses while the completion report claims all records were embedded. External models are dependencies that need version governance and compatibility testing. After endpoint upgrade structured responses no longer satisfy the application's JSON schema.

B:

The same document produces different field names and invalid JSON across retries. External models are dependencies that need version governance and compatibility testing. After endpoint upgrade structured responses no longer satisfy the application's JSON schema.

C:

The secret appears in developer-readable config and can invoke the endpoint outside the approved process. External models are dependencies that need version governance and compatibility testing. After endpoint upgrade structured responses no longer satisfy the application's JSON schema.

Question 270

A SQL deployment runs a stored procedure that seeds lookup codes and each replay inserts duplicate reference rows. What would be a valid post-database-change verification?

  1. Running the same deployment twice produces the expected reference rows without duplication.
  2. A clean rebuild and incremental deployment produce identical authorized lookup mappings.
  3. Tests return the same correctly scoped rows for all clients without a substantial plan regression.
  4. Unauthorized selects are blocked by permissions and legitimate data changes produce reliable audit records.

Explanation

Correct Answer

A:

Running the same deployment twice produces the expected reference rows without duplication.

Incorrect Answers

B:

A clean rebuild and incremental deployment produce identical authorized lookup mappings. Running the same deployment twice produces the expected reference rows without duplication.

C:

Tests return the same correctly scoped rows for all clients without a substantial plan regression. Running the same deployment twice produces the expected reference rows without duplication.

D:

Unauthorized selects are blocked by permissions and legitimate data changes produce reliable audit records. Running the same deployment twice produces the expected reference rows without duplication.

Developing AI-Enabled Database Solutions (DP-800) Practice Tests