Question 250
A hosted inference integration needs independently revocable Azure permissions that are not tied to an engineer's personal login. Which option would you recommend?
- Database role membership
- Personal developer token
- Group-based database access
- Dedicated managed identity
Explanation
Correct Answer
D:A dedicated managed identity offers a workload principal whose token authorization can be controlled independently of human account lifecycle.
Incorrect Answers
A:A database role collects permissions for members and reduces the need to duplicate identical grants on individual principals; the missing behavior is Dedicated managed identity; this mechanism does something else.
B:A personal developer token ties service access to an individual and may require rotation if that person leaves; it is not a separate revocable workload identity.
C:Group-backed principals allow permissions to be managed through authorized group membership while database grants remain scoped; for the stated conditions, the deciding feature is Dedicated managed identity rather than this behavior.
Question 251
A team needs TWO deliverable artifact types or project formats, not a command or deployment operation: the declarative, cross-platform project format and the schema-only built package that release automation can publish. Choose TWO.
- DACPAC
- `dotnet build` of SQL project
- SqlPackage Publish
- BACPAC
- SDK-style SQL database project
Explanation
Correct Answers
A:A DACPAC packages the database schema model for DacFx comparison and deployment; it is not primarily a full data export; used in combination with SDK-style SQL database project, it handles a separate requirement of this combined workflow.
E:An SDK-style SQL project provides declarative schema files and cross-platform build tooling supported by Microsoft.Build.Sql; used in combination with DACPAC, it handles a separate requirement of this combined workflow.
Incorrect Answers
B:Building an SDK-style SQL project validates its schema model and emits a DACPAC when the build succeeds; the alternative cannot substitute for DACPAC and SDK-style SQL database project under the given constraints.
C:SqlPackage Publish compares the source package with the destination and executes a generated deployment plan when authorized; the condition in the stem is resolved by DACPAC and SDK-style SQL database project, not by this feature alone.
D:A BACPAC is a logical export containing schema and data, unlike a schema-only DACPAC; the missing behavior is DACPAC and SDK-style SQL database project; this mechanism does something else.
Question 252
A hosted REST API should accept enterprise-issued bearer identities and validate their claims before mapping them to DAB roles. What should be configured to meet this objective?
- DAB authorization roles
- TLS ingress
- Database Row-Level Security
- Entra JWT authentication
Explanation
Correct Answer
D:An Entra-compatible authentication provider validates signed tokens for a configured audience and issuer before authorization.
Incorrect Answers
A:DAB entity role permissions limit which API operations authenticated roles can invoke; for the stated conditions, the deciding feature is Entra JWT authentication rather than this behavior.
B:HTTPS/TLS protects requests and responses in transit while leaving entity permissions and SQL grants to separate controls; it addresses a different step; the question calls for Entra JWT authentication.
C:SQL Row-Level Security applies within the database and complements, rather than replaces, API operation permissions; this differs from the Entra JWT authentication requirement described in the scenario.
Question 253
A deployment can use native VECTOR storage and exact distance calculations even when its engine does not enable preview ANN indexing. Which choice best meets these constraints?
- Azure SQL generally available vector index
- Version-specific syntax review
- Vector candidate pipeline
- SQL Server 2025 preview indexing
- Exact SQL vector functions
Explanation
Correct Answer
E:VECTOR and VECTOR_DISTANCE provide exact vector operations separately from optional approximate index support.
Incorrect Answers
A:Vector indexes and approximate VECTOR_SEARCH are generally available in Azure SQL Database, with version-specific syntax requirements; the condition in the stem is resolved by Exact SQL vector functions, not by this feature alone.
B:Current documentation favors SELECT TOP (N) WITH APPROXIMATE for modern indexes, while older TOP_N syntax remains a compatibility concern; the missing behavior is Exact SQL vector functions; this mechanism does something else.
C:Vector-neighbor retrieval produces semantic candidate IDs and distance-based order for the second fusion input; that function is not interchangeable with Exact SQL vector functions for the requested result.
D:On SQL Server 2025, vector indexing and VECTOR_SEARCH have preview conditions that differ from generally available Azure SQL support; for the stated conditions, the deciding feature is Exact SQL vector functions rather than this behavior.
Question 254
Search results must first be restricted to the authenticated tenant and current language before relevance is considered. What is the most appropriate design decision?
- Exact nearest-neighbor scan
- Keyword inverted index
- Approximate nearest-neighbor index
- Metadata predicate pushdown
Explanation
Correct Answer
D:Applying supported metadata predicates early reduces candidate sets and enforces eligibility independently of semantic ranking.
Incorrect Answers
A:Exact nearest-neighbor search computes distances across all eligible candidates and returns the true nearest results for the chosen metric; this differs from the Metadata predicate pushdown requirement described in the scenario.
B:An inverted full-text index maps terms to matching documents for lexical search over tokenized content; the required capability here is Metadata predicate pushdown, which this operation does not supply.
C:Approximate-neighbor indexes reduce the amount of distance work by searching a structured vector index, trading some recall for speed; the alternative cannot substitute for Metadata predicate pushdown under the given constraints.
Question 255
Imported date strings may be malformed; invalid conversions should return NULL instead of aborting a query. Which database mechanism should be chosen?
- CASE expression
- COALESCE
- TRY_CONVERT
- NULLIF
Explanation
Correct Answer
C:TRY_CONVERT attempts type conversion and returns NULL for many invalid values where a regular CONVERT would fail.
Incorrect Answers
A:CASE evaluates conditional branches and yields a scalar expression selected from matching WHEN conditions; that function is not interchangeable with TRY_CONVERT for the requested result.
B:COALESCE returns the first non-NULL expression from its ordered arguments; for the stated conditions, the deciding feature is TRY_CONVERT rather than this behavior.
D:NULLIF returns NULL when its two expressions are equal, often used with division to handle zero denominators; this differs from the TRY_CONVERT requirement described in the scenario.
Question 256
A frequently requested read-only catalog query needs short-lived response caching in the API runtime to reduce repeated SQL calls. Which option satisfies the desired behavior most directly?
- DAB Redis cache
- DAB in-memory response cache
- `$first` page-size control
- `dab start`
- GraphQL pagination `after`
Explanation
Correct Answer
B:DAB response caching can retain eligible read results for configured durations, subject to invalidation and consistency requirements.
Incorrect Answers
A:A supported distributed Redis cache stores response entries outside individual runtime processes for multi-instance reuse; the alternative cannot substitute for DAB in-memory response cache under the given constraints.
C:The first parameter requests the page size; it does not itself identify where the next page begins; the missing behavior is DAB in-memory response cache; this mechanism does something else.
D:The start command runs the configured DAB service and hosts enabled endpoints for requests; that function is not interchangeable with DAB in-memory response cache for the requested result.
E:A GraphQL after cursor continues pagination beyond a prior page, subject to the configured query shape; for the stated conditions, the deciding feature is DAB in-memory response cache rather than this behavior.
Question 257
A reporting query must read source records that remain in an external system represented through a SQL external table definition. What is the right solution for this specific requirement?
- Ledger verification
- FOR SYSTEM_TIME AS OF
- Graph MATCH predicate
- External data access
Explanation
Correct Answer
D:An external table exposes remote data through metadata and suitable connectivity without copying it into a regular local table first.
Incorrect Answers
A:Ledger verification uses generated digests and recorded history to detect tampering; it is not simply a time-travel query; it addresses a different step; the question calls for External data access.
B:FOR SYSTEM_TIME AS OF retrieves the temporal version valid at a requested instant using the system-maintained period columns; this differs from the External data access requirement described in the scenario.
C:MATCH expresses a graph traversal pattern over node and edge tables rather than a plain scalar comparison; the required capability here is External data access, which this operation does not supply.
Question 258
A legal workflow primarily needs to find the exact paragraph containing a statutory citation, not generate a synthesized answer. Which action is justified by these conditions?
- Fine-tuning
- SQL structured lookup
- Full-text document retrieval
- Retrieval-augmented generation
Explanation
Correct Answer
C:Full-text search can locate matching passages directly; generation adds complexity when the task is simply retrieval.
Incorrect Answers
A:Fine-tuning adjusts model behavior through training on examples; it is not the usual replacement for retrieving frequently changing knowledge; the missing behavior is Full-text document retrieval; this mechanism does something else.
B:A parameterized structured SQL lookup provides an exact current record without relying on probabilistic text generation; that function is not interchangeable with Full-text document retrieval for the requested result.
D:RAG retrieves relevant current documents and provides them as context to the generator, keeping knowledge retrieval distinct from model weights; it addresses a different step; the question calls for Full-text document retrieval.
Question 259
A read-only API principal must be unable to UPDATE the Orders table even if it can run the approved read procedure. Which action is justified by these conditions?
- Positive-path unit test
- Boundary-value test
- Permission-negative test
- Migration integration test
Explanation
Correct Answer
C:Negative authorization tests verify forbidden operations fail with the actual application principal rather than assuming intended grants.
Incorrect Answers
A:Positive-path unit tests validate intended output and state changes against known inputs; the required capability here is Permission-negative test, which this operation does not supply.
B:Boundary tests target conditions at and near validation thresholds where logic defects commonly occur; the alternative cannot substitute for Permission-negative test under the given constraints.
D:Integration tests after migration exercise live data transformations, object dependencies, and application-critical behavior; the missing behavior is Permission-negative test; this mechanism does something else.
Question 260
Vector results are slow because almost all rows belong to ineligible tenants; engineers need to evaluate filtering strategy and candidate sizes. Which option satisfies the desired behavior most directly?
- ANN index evaluation
- Distance-computation cost
- Filter-selectivity assessment
- Top-k selection
Explanation
Correct Answer
C:Selective eligibility predicates and supported filter execution behavior affect vector retrieval work and correctness.
Incorrect Answers
A:ANN indexing can reduce comparisons and improve speed, subject to measured recall and platform support; for the stated conditions, the deciding feature is Filter-selectivity assessment rather than this behavior.
B:Exhaustive distance calculations scale with candidate count and vector dimension, creating CPU and sorting pressure; it addresses a different step; the question calls for Filter-selectivity assessment.
D:Limiting k reduces result volume and downstream processing while preserving the intended nearest-neighbor objective; the required capability here is Filter-selectivity assessment, which this operation does not supply.