Microsoft DP-800 Questions & Answers

Parsing nested arrays into typed rows reduces fragile manual string extraction. The actual technical constraint is that unindexed JSON path predicates can incur expensive row-by-row parsing.

Sequence numbers can be consumed even when transactions roll back or cache is lost. The actual technical constraint is that unindexed JSON path predicates can incur expensive row-by-row parsing.

Finite ranges exhaust and enabling CYCLE without safety checks may repeat identifiers. The actual technical constraint is that unindexed JSON path predicates can incur expensive row-by-row parsing.

Alert on per-document cost and failed queue age. The current task instead requires this action: audit query and corpus model IDs on every release.

Reject unexpected fields and never trust generated totals without validation.

Handle replay, deletion and watermark recovery. The current task instead requires this action: audit query and corpus model IDs on every release.

A path existence predicate does not necessarily compute minimal hop count.

NULL in a NOT IN subquery can yield UNKNOWN and unexpectedly filter all rows.

Transient retries can replay committed work even after client-side timeouts.

The computed amount matches audited ledger SQL exactly. However, that result would not establish the required outcome: a cross-client negative test yields no unauthorized quoted text.

A JSON parser reconstructs the intended order IDs and amounts. However, that result would not establish the required outcome: a cross-client negative test yields no unauthorized quoted text.

The approved identity receives a 2xx response while an unauthorized one fails. However, that result would not establish the required outcome: a cross-client negative test yields no unauthorized quoted text.

JSON_VALUE extracts a scalar path, while OPENJSON can shred collections. The actual technical constraint is that JSON_ARRAYAGG can collect multiple row values into a JSON array.

LAG references a preceding ordered row without self-joining every record. The actual technical constraint is that JSON_ARRAYAGG can collect multiple row values into a JSON array.

An outer reference lets the inner aggregate vary by each row's category. The actual technical constraint is that JSON_ARRAYAGG can collect multiple row values into a JSON array.

Run tests in CI with controlled representative fixtures. The current task instead requires this action: keep isolated test credentials and avoid production data.

Enforce secret masking and log retention policy. The current task instead requires this action: keep isolated test credentials and avoid production data.

Define drift policy for emergency fixes. The current task instead requires this action: keep isolated test credentials and avoid production data.

Audit collection adds actor, operation and time evidence but does not itself block changes.

Read filtering alone does not necessarily stop unauthorized DML that changes row ownership.

This action addresses a different case: An RLS policy uses SESSION_CONTEXT set directly from untrusted request text. For that case, session context is only safe as an authorization input when callers cannot impersonate other tenants. For this case, it does not directly correct the observed problem; the distinguishing issue is high-volume blanket logging can impact throughput and reveal sensitive inputs. The specifically relevant immediate technical action is: Use targeted audit groups and evaluate performance and retention costs.

The app reads the intended view but cannot read private financial objects. However, that result would not establish the required outcome: the worker connects with an approved identity and no stored SQL password.

The legitimate procedure succeeds while unrelated objects remain inaccessible. However, that result would not establish the required outcome: the worker connects with an approved identity and no stored SQL password.

A spoofed request cannot change access to another tenant's rows. However, that result would not establish the required outcome: the worker connects with an approved identity and no stored SQL password.

Levenshtein-style distance measures character edits, not verified real-world identity. The actual technical constraint is that jaro-Winkler-style measures can treat transpositions and prefixes differently from raw edits.

Without coordinated error handling, partial writes can violate business invariants. The actual technical constraint is that jaro-Winkler-style measures can treat transpositions and prefixes differently from raw edits.

Silently suppressed exceptions create false evidence of committed work. The actual technical constraint is that jaro-Winkler-style measures can treat transpositions and prefixes differently from raw edits.

Keep section IDs and overlap rules versioned. The current task instead requires this action: review boundary cases during chunker changes.

Audit query and corpus model IDs on every release. The current task instead requires this action: review boundary cases during chunker changes.

Inspect invocation logs and rate limits. The current task instead requires this action: review boundary cases during chunker changes.

Model invocation can export inputs and incur charges, so object privileges matter.

Embedding inputs can expose or encode sensitive data without adding retrieval value.

Text-only models cannot directly interpret image content without a supporting vision pipeline.