McAfee Secure

Exam Code: DP-800

Exam Name: Developing AI-Enabled Database Solutions

Certification Provider: Microsoft

Corresponding Certification: Microsoft Certified: SQL AI Developer Associate

Questions & Answers

Microsoft DP-800 Questions & Answers

Study with Up-To-Date REAL Exam Questions and Answers from the ACTUAL Test

Questions & Answers

101 Questions & Answers with Testing Engine

"Developing AI-Enabled Database Solutions Exam", also known as DP-800 exam, is a Microsoft certification exam.

Pass your tests with the always up-to-date DP-800 Exam Engine. Your DP-800 training materials keep you at the head of the pack!

Satisfaction Guaranteed

Money Back Guarantee

Testking provides hassle-free money back guarantee with our products. That is because we have 100% trust in the abilities of our professional and experience product team, and our record is a proof of that.

99.6% PASS RATE
Was: $137.49
Now: $124.99

Microsoft DP-800 Practice Test 15

Question 140

An API expects a single relational row converted into a JSON object with named fields. Following remediation, what safeguard BEST prevents a repeat of this specific failure?

  1. Index frequently filtered paths only after measured need.
  2. Keep review logs and a reversible merge mechanism.
  3. Use selective indexes on the correlated join key.
  4. Pin feature availability to the deployed SQL engine.

Explanation

Correct Answer

D:

Pin feature availability to the deployed SQL engine. This guards against recurrence because JSON_OBJECT constructs a JSON object from typed expressions instead of fragile string concatenation. The direct remediation is: Use JSON_OBJECT with explicit key/value expressions where supported.

Incorrect Answers

A:

Index frequently filtered paths only after measured need. The current task instead requires this action: pin feature availability to the deployed SQL engine.

B:

Keep review logs and a reversible merge mechanism. The current task instead requires this action: pin feature availability to the deployed SQL engine.

C:

Use selective indexes on the correlated join key. The current task instead requires this action: pin feature availability to the deployed SQL engine.

Question 141

A DBA hotfix changes a production column outside source control and the next publish plans to reverse it. Which technical action BEST resolves the primary problem?

  1. Compare live database schema against the project model before approval.
  2. Run schema comparison or drift reporting against the target environment.
  3. Review migration impact, data retention and dependent object behavior before approval.
  4. Add integration tests across database permissions, DAB configuration and HTTP behavior.

Explanation

Correct Answer

A:

Compare live database schema against the project model before approval. This targets the primary issue because manual changes create drift between intended declarative source and deployed objects. Verification: Drift detection flags the unexpected column change before publish. Prevent recurrence: Reconcile sanctioned hotfixes into source control promptly.

Incorrect Answers

B:

Running schema comparison or drift reporting against the target environment addresses a different database behavior; the relevant distinction is that compare live database schema against the project model before approval.

C:

Reviewing migration impact, data retention and dependent object behavior before approval addresses a different database behavior; the relevant distinction is that compare live database schema against the project model before approval.

D:

Cross-system contracts can fail even when individual SQL procedures are correct.

Question 142

An administrator overrides a failed security scan because pipeline approval rules are not mandatory. Which test result BEST demonstrates that the intended solution is working?

  1. A billing SQL PR requires a finance-code-owner approval before merge.
  2. A test release with a failing security check is blocked.
  3. The merged procedure passes both teams' regression tests.
  4. The pipeline reaches only its intended staging database and resource group.

Explanation

Correct Answer

B:

A test release with a failing security check is blocked. This is the decisive observable evidence after implementing make required test/security checks and approval gates nonbypassable by normal contributors. It tests the underlying requirement: Optional checks fail to enforce controls during urgent releases.

Incorrect Answers

A:

A billing SQL PR requires a finance-code-owner approval before merge. However, that result would not establish the required outcome: a test release with a failing security check is blocked.

C:

The merged procedure passes both teams' regression tests. However, that result would not establish the required outcome: a test release with a failing security check is blocked.

D:

The pipeline reaches only its intended staging database and resource group. However, that result would not establish the required outcome: a test release with a failing security check is blocked.

Question 143

A team enables every MCP tool and finds the assistant can execute potentially destructive SQL. Which technical principle BEST explains the relevant design decision?

  1. Different model endpoints can have different data governance and retention arrangements.
  2. Personal and enterprise assistants can have different retention and policy controls.
  3. Broad MCP tool capability expands the agent's potential effect on the database.
  4. A malicious or compromised MCP server may exfiltrate data or supply unsafe tool behavior.

Explanation

Correct Answer

C:

Broad MCP tool capability expands the agent's potential effect on the database. This explains why the corrective action is to reduce tool permissions to narrowly scoped tasks and require human confirmation for writes. Confirm the diagnosis when a dry run refuses dangerous actions but permits approved diagnostics.

Incorrect Answers

A:

Different model endpoints can have different data governance and retention arrangements. The actual technical constraint is that broad MCP tool capability expands the agent's potential effect on the database.

B:

Personal and enterprise assistants can have different retention and policy controls. The actual technical constraint is that broad MCP tool capability expands the agent's potential effect on the database.

D:

A malicious or compromised MCP server may exfiltrate data or supply unsafe tool behavior. The actual technical constraint is that broad MCP tool capability expands the agent's potential effect on the database.

Question 144

A search team will retune its ANN vector index monthly. The index is fast, but unnoticed drops in relevant-product recall would harm customers. What preventive quality practice should accompany each release?

  1. Record only the p95 query latency and approve any change that reduces it.
  2. Discard labeled search queries after building each new index.
  3. Assume a successful index creation proves equivalent results to exhaustive search.
  4. Maintain a versioned exact-KNN benchmark with labeled queries and compare ANN recall@K and latency after each index change.

Explanation

Correct Answer

D:

An exact baseline detects recall loss while a separate latency threshold checks whether ANN still meets its performance objective.

Incorrect Answers

A:

ANN can become faster while omitting relevant neighbors; latency alone is an incomplete quality gate.

B:

Without a stable reference set the team cannot measure whether retrieval quality regresses between releases.

C:

Building an approximate index does not demonstrate that its neighbors match an exact-distance baseline.

Question 145

A Fabric engineer expects Copilot SQL help in a tenant where capacity and admin settings are disabled. Which technical action BEST resolves the primary problem?

  1. Review Fabric capacity eligibility and tenant Copilot admin controls before testing.
  2. Confirm approved GitHub Copilot seat, editor extension and organization policy.
  3. Keep static SQL checks, pull-request review and least-privilege deployment despite instructions.
  4. Select an organization-approved model and review data-routing settings before sending schema text.

Explanation

Correct Answer

A:

Review Fabric capacity eligibility and tenant Copilot admin controls before testing. This targets the primary issue because copilot in Fabric availability can depend on region, capacity and tenant policies. Verification: An approved SQL workspace exposes supported Copilot features for a licensed tester. Prevent recurrence: Revalidate Fabric feature availability when capacity policies change.

Incorrect Answers

B:

Editor AI availability depends on licensing, identity and permitted configuration.

C:

Prompt instructions influence suggestions but are not mandatory security policies.

D:

Different model endpoints can have different data governance and retention arrangements.

Question 146

An ingestion job ignores rate limits and retries a costly embedding API in a tight loop. Which test result BEST demonstrates that the intended solution is working?

  1. All new embeddings insert without dimension errors and search quality is rechecked.
  2. A throttled test recovers without duplicate vector records or budget spikes.
  3. An authorized test generates an embedding with the expected size.
  4. A sampled outbound model request contains no restricted secrets.

Explanation

Correct Answer

B:

A throttled test recovers without duplicate vector records or budget spikes. This is the decisive observable evidence after implementing use bounded exponential backoff, batching where supported and idempotent writes. It tests the underlying requirement: External model quotas and transient errors require controlled retry behavior.

Incorrect Answers

A:

All new embeddings insert without dimension errors and search quality is rechecked. However, that result would not establish the required outcome: a throttled test recovers without duplicate vector records or budget spikes.

C:

An authorized test generates an embedding with the expected size. However, that result would not establish the required outcome: a throttled test recovers without duplicate vector records or budget spikes.

D:

A sampled outbound model request contains no restricted secrets. However, that result would not establish the required outcome: a throttled test recovers without duplicate vector records or budget spikes.

Question 147

Child invoice details remain after their parent invoices are removed unexpectedly. What is the MOST relevant technical distinction in this situation?

  1. Application-layer validation alone can be bypassed by other writers.
  2. A sequence can allocate numbers independently of one table identity property.
  3. Referential integrity prevents orphan children unless cascade or other actions are deliberately configured.
  4. Sequence numbers can be consumed even when transactions roll back or cache is lost.

Explanation

Correct Answer

C:

Referential integrity prevents orphan children unless cascade or other actions are deliberately configured. This explains why the corrective action is to define and test a FOREIGN KEY and explicitly select appropriate delete behavior. Confirm the diagnosis when orphan inserts fail and approved delete scenarios behave as specified.

Incorrect Answers

A:

Application-layer validation alone can be bypassed by other writers. The actual technical constraint is that referential integrity prevents orphan children unless cascade or other actions are deliberately configured.

B:

A sequence can allocate numbers independently of one table identity property. The actual technical constraint is that referential integrity prevents orphan children unless cascade or other actions are deliberately configured.

D:

Sequence numbers can be consumed even when transactions roll back or cache is lost. The actual technical constraint is that referential integrity prevents orphan children unless cascade or other actions are deliberately configured.

Question 148

An imported address field contains an embedded postal-code pattern and needs extraction. Which ongoing control is MOST relevant after the issue is fixed?

  1. Check engine support and trim blank tokens before inserting.
  2. Use deterministic tie-breakers in the ORDER BY.
  3. Pin feature availability to the deployed SQL engine.
  4. Validate format differences across target geographies.

Explanation

Correct Answer

D:

Validate format differences across target geographies. This guards against recurrence because regex substring extraction returns the pattern occurrence, not a tokenized table. The direct remediation is: Use REGEXP_SUBSTR or an equivalent supported regex extraction function.

Incorrect Answers

A:

Check engine support and trim blank tokens before inserting. The current task instead requires this action: validate format differences across target geographies.

B:

Use deterministic tie-breakers in the ORDER BY. The current task instead requires this action: validate format differences across target geographies.

C:

Pin feature availability to the deployed SQL engine. The current task instead requires this action: validate format differences across target geographies.

Question 149

A migration switches embedding models from 768 to 1536 dimensions without changing the SQL schema. What is the MOST appropriate immediate corrective approach?

  1. Plan a new compatible vector column/index and re-embed the corpus using the new model.
  2. Use bounded exponential backoff, batching where supported and idempotent writes.
  3. Move model inference out of the trigger and use a durable change notification pipeline.
  4. Keep rapidly changing structured quantities as filterable columns outside the descriptive embedding text.

Explanation

Correct Answer

A:

Plan a new compatible vector column/index and re-embed the corpus using the new model. This targets the primary issue because vector dimensionality is fixed by the column and must match the query vector. Verification: All new embeddings insert without dimension errors and search quality is rechecked. Prevent recurrence: Version model identity with stored vectors and avoid mixed spaces.

Incorrect Answers

B:

External model quotas and transient errors require controlled retry behavior.

C:

Remote inference inside DML triggers lengthens database transactions and increases failure coupling.

D:

Frequent numeric fields add update cost and can weaken semantic relevance.

Question 150

An Azure SQL table contains textual cases but the vector column is always NULL. What observed evidence would BEST confirm successful remediation?

  1. A throttled test recovers without duplicate vector records or budget spikes.
  2. Test rows contain supported dimension vectors and corresponding text IDs.
  3. A bilingual test set retrieves correct permitted items for each locale.
  4. Updated products receive new embeddings within the freshness SLA.

Explanation

Correct Answer

B:

Test rows contain supported dimension vectors and corresponding text IDs. This is the decisive observable evidence after implementing call the approved embedding model through a supported integration and store fixed-dimension results. It tests the underlying requirement: Embeddings are computed numeric representations and are not created merely by defining a VECTOR column.

Incorrect Answers

A:

A throttled test recovers without duplicate vector records or budget spikes. However, that result would not establish the required outcome: test rows contain supported dimension vectors and corresponding text IDs.

C:

A bilingual test set retrieves correct permitted items for each locale. However, that result would not establish the required outcome: test rows contain supported dimension vectors and corresponding text IDs.

D:

Updated products receive new embeddings within the freshness SLA. However, that result would not establish the required outcome: test rows contain supported dimension vectors and corresponding text IDs.

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