Microsoft DP-900 Practice Test 72
Question 710
A lakehouse stores raw sensor payloads cleaned records and business summary tables. After selecting the proper Bronze silver gold medallion design approach, which observation gives the clearest evidence of success?
- Gold measures reconcile to corrected silver and source bronze records.
- Intended filtered events appear at the configured downstream consumers.
- Images remain retrievable while SQL reports query approved modeled facts.
- Filtered aggregations match approved transactional totals.
Explanation
Correct Answer
A:Gold measures reconcile to corrected silver and source bronze records. This is the directly observable result for Bronze silver gold medallion design; medallion layers make data refinement and reuse explicit.
Incorrect Answers
B:Intended filtered events appear at the configured downstream consumers. This supports Fabric Eventstreams real-time routing, not confirmation of Bronze silver gold medallion design. The relevant evidence is that gold measures reconcile to corrected silver and source bronze records.
C:Images remain retrievable while SQL reports query approved modeled facts. This supports Structured warehouse and unstructured lake coexistence, not confirmation of Bronze silver gold medallion design. The relevant evidence is that gold measures reconcile to corrected silver and source bronze records.
D:Filtered aggregations match approved transactional totals. This supports Star schema report model, not confirmation of Bronze silver gold medallion design. The relevant evidence is that gold measures reconcile to corrected silver and source bronze records.
Question 711
A nightly sales report can tolerate a daily scheduled import. What is the MOST appropriate next validation or correction?
- Call a once-daily CSV read a continuous real-time stream.
- Measure event-to-alert time and completion times separately.
- Test one authorized and one revoked device connection.
- Test hierarchy navigation and chronological sorting.
Explanation
Correct Answer
B:Measure event-to-alert time and completion times separately. This tests the intended result: fraud alerts arrive promptly while daily reports complete within their SLA.
Incorrect Answers
A:Call a once-daily CSV read a continuous real-time stream. This unsafe shortcut leaves the Streaming versus batch latency choice problem unresolved. Running hourly batch for urgent alerts creates unacceptable delay.
C:Test one authorized and one revoked device connection. It tests Azure IoT Hub device messaging rather than the critical Streaming versus batch latency choice behavior. Streaming processes arriving records continually unlike scheduled batch.
D:Test hierarchy navigation and chronological sorting. It tests Drill-down hierarchy for time rather than the critical Streaming versus batch latency choice behavior. Streaming processes arriving records continually unlike scheduled batch.
Question 712
A table has a surrogate primary key but needs a separate unique tax identifier. What is the MOST appropriate next validation or correction?
- Test missing-parent inserts and deletion cases under the actual constraint.
- Test simultaneous duplicate inserts and database errors.
- Map required features to each documented service model.
- Put UNIQUE in a comment and assume the database validates it.
Explanation
Correct Answer
B:Test simultaneous duplicate inserts and database errors. This tests the intended result: duplicate tax identifiers are rejected without changing row primary keys.
Incorrect Answers
A:Test missing-parent inserts and deletion cases under the actual constraint. It tests Foreign key referential integrity rather than the critical Unique index versus primary key behavior. A table can enforce uniqueness on nonprimary attributes.
C:Map required features to each documented service model. It tests SQL deployment choice by control plane rather than the critical Unique index versus primary key behavior. A table can enforce uniqueness on nonprimary attributes.
D:Put UNIQUE in a comment and assume the database validates it. This unsafe shortcut leaves the Unique index versus primary key problem unresolved. Relying solely on application checks allows races that create duplicates.
Question 713
A vendor assumes Azure patches its stored procedures and application drivers. Which choice BEST meets this requirement?
- Choose Azure Database for PostgreSQL Flexible Server for supported Postgres workloads.
- Separate provider engine/host operations from customer schema data and application responsibilities.
- Assume subscribing to PaaS automatically fixes every bad SQL query.
- Match aggregation grain and relationship cardinality before summing measures.
Explanation
Correct Answer
B:PaaS reduces infrastructure upkeep but does not remove all application DBA duties.
Incorrect Answers
A:Choose Azure Database for PostgreSQL Flexible Server for supported Postgres workloads. That action primarily handles Azure Database for PostgreSQL Flexible Server; the given requirement is Azure SQL managed maintenance responsibility. PaaS reduces infrastructure upkeep but does not remove all application DBA duties.
C:Assume subscribing to PaaS automatically fixes every bad SQL query. It does not implement Azure SQL managed maintenance responsibility. Neglecting schema access reviews can leave insecure data even on patched engines.
D:Match aggregation grain and relationship cardinality before summing measures. That action primarily handles Cardinality of joined dimensions; the given requirement is Azure SQL managed maintenance responsibility. PaaS reduces infrastructure upkeep but does not remove all application DBA duties.
Question 714
A retailer compares revenue across ten product families. After selecting the proper Column chart category comparison approach, which observation gives the clearest evidence of success?
- Relative category sizes and labels are clear.
- Supported Spark and SQL experiences access the intended lakehouse tables.
- Points reveal clustering and possible outliers without asserting causality.
- Consumers read partitioned event streams with tracked sequence and retention.
Explanation
Correct Answer
A:Relative category sizes and labels are clear. This is the directly observable result for Column chart category comparison; bar and column charts support comparisons across discrete categories.
Incorrect Answers
B:Supported Spark and SQL experiences access the intended lakehouse tables. This supports Fabric Lakehouse managed tables and files, not confirmation of Column chart category comparison. The relevant evidence is that relative category sizes and labels are clear.
C:Points reveal clustering and possible outliers without asserting causality. This supports Scatter chart relationship, not confirmation of Column chart category comparison. The relevant evidence is that relative category sizes and labels are clear.
D:Consumers read partitioned event streams with tracked sequence and retention. This supports Azure Event Hubs event ingestion, not confirmation of Column chart category comparison. The relevant evidence is that relative category sizes and labels are clear.
Question 715
A business needs a secondary-region SQL database for disaster recovery. Which choice BEST meets this requirement?
- Choose Azure SQL Database and size the individual database compute tier.
- Use INNER JOIN on the intended keys when only matching pairs are required.
- Choose supported Azure SQL geo-replication or failover groups with realistic RPO/RTO.
- Assume a second region stores every last write instantaneously in every deployment model.
Explanation
Correct Answer
C:Geo-replicated SQL designs may use asynchronous replication and have possible lag.
Incorrect Answers
A:Choose Azure SQL Database and size the individual database compute tier. That action primarily handles Azure SQL Database single database PaaS; the given requirement is Azure SQL active geo-replication tradeoff. Geo-replicated SQL designs may use asynchronous replication and have possible lag.
B:Use INNER JOIN on the intended keys when only matching pairs are required. That action primarily handles Inner join row matching; the given requirement is Azure SQL active geo-replication tradeoff. Geo-replicated SQL designs may use asynchronous replication and have possible lag.
D:Assume a second region stores every last write instantaneously in every deployment model. It does not implement Azure SQL active geo-replication tradeoff. Ignoring lag could lose recent writes at regional failover.
Question 716
An operations app repeatedly applies the same approved order status transition. If the relevant Stored procedure reusable server operation safeguard is skipped, which failure is MOST directly associated with that omission?
- Treating MI like a random public VM port without subnet planning can block migration.
- Selecting ungrouped detail fields can produce invalid or misleading results.
- A lift-and-shift that ignores dependencies can fail despite successful bulk load.
- Unvalidated concatenation inside procedures can still create injection risk.
Explanation
Correct Answer
D:Unvalidated concatenation inside procedures can still create injection risk. Procedures encapsulate server-side logic separate from simple SELECT views. A suitable check is to test parameter validation and execution permissions.
Incorrect Answers
A:Treating MI like a random public VM port without subnet planning can block migration. This failure is associated with SQL Managed Instance virtual network; it is not the primary failure in the stated Stored procedure reusable server operation scenario. Procedures encapsulate server-side logic separate from simple SELECT views.
B:Selecting ungrouped detail fields can produce invalid or misleading results. This failure is associated with SQL GROUP BY aggregate; it is not the primary failure in the stated Stored procedure reusable server operation scenario. Procedures encapsulate server-side logic separate from simple SELECT views.
C:A lift-and-shift that ignores dependencies can fail despite successful bulk load. This failure is associated with Data migration SQL Server assessment; it is not the primary failure in the stated Stored procedure reusable server operation scenario. Procedures encapsulate server-side logic separate from simple SELECT views.
Question 717
A registration pipeline is missing addresses for one country. If the relevant Data quality completeness and accuracy safeguard is skipped, which failure is MOST directly associated with that omission?
- Treating production checkout row updates as a warehouse's primary workload can hurt analytics design.
- A pipeline that merely completes does not prove data correctness.
- Replacing old dimension values can rewrite historical business interpretation.
- Expecting server-side joins and foreign keys in Table Storage fails.
Explanation
Correct Answer
B:A pipeline that merely completes does not prove data correctness. Data quality requires explicit rules tied to business meaning. A suitable check is to reconcile known records and expected missing-value rates.
Incorrect Answers
A:Treating production checkout row updates as a warehouse's primary workload can hurt analytics design. This failure is associated with Analytical warehouse for structured reporting; it is not the primary failure in the stated Data quality completeness and accuracy scenario. Data quality requires explicit rules tied to business meaning.
C:Replacing old dimension values can rewrite historical business interpretation. This failure is associated with Historical dimensions versus current operational truth; it is not the primary failure in the stated Data quality completeness and accuracy scenario. Data quality requires explicit rules tied to business meaning.
D:Expecting server-side joins and foreign keys in Table Storage fails. This failure is associated with Azure Table Storage key-value entities; it is not the primary failure in the stated Data quality completeness and accuracy scenario. Data quality requires explicit rules tied to business meaning.
Question 718
A vendor needs temporary read-only access to one report blob. Which choice BEST meets this requirement?
- Configure supported Azure Files snapshot or backup protection.
- Use Azure Cosmos DB for NoSQL with suitable documents and container partition strategy.
- Issue a limited SAS or supported delegated access scoped by operation resource and expiry.
- Publish the full account access key in a URL and call it an expiring read-only SAS.
Explanation
Correct Answer
C:Shared access signatures grant bounded access without broad permanent account credentials.
Incorrect Answers
A:Configure supported Azure Files snapshot or backup protection. That action primarily handles Azure Files snapshots and backups; the given requirement is Blob SAS delegated access. Shared access signatures grant bounded access without broad permanent account credentials.
B:Use Azure Cosmos DB for NoSQL with suitable documents and container partition strategy. That action primarily handles Cosmos DB NoSQL JSON document container; the given requirement is Blob SAS delegated access. Shared access signatures grant bounded access without broad permanent account credentials.
D:Publish the full account access key in a URL and call it an expiring read-only SAS. It does not implement Blob SAS delegated access. Overly long or permissive SAS tokens expose data if forwarded.
Question 719
A published report still shows last week's sales due to a failed refresh. What is the MOST appropriate next validation or correction?
- Re-run transformations with new sample rows.
- Close the desktop laptop and assume private SQL data continues refreshing magically.
- Inspect gateway connection refresh history and credential state.
- Review model relationships and test grouped business metrics.
Explanation
Correct Answer
C:Inspect gateway connection refresh history and credential state. This tests the intended result: cloud refresh succeeds with secured access to the intended on-premises source.
Incorrect Answers
A:Re-run transformations with new sample rows. It tests Power Query shaping transformation rather than the critical Power BI scheduled refresh and gateway behavior. Gateways bridge supported private sources for managed cloud refresh workflows.
B:Close the desktop laptop and assume private SQL data continues refreshing magically. This unsafe shortcut leaves the Power BI scheduled refresh and gateway problem unresolved. Publishing a report does not guarantee the local source is reachable.
D:Review model relationships and test grouped business metrics. It tests Star schema analytical query model rather than the critical Power BI scheduled refresh and gateway behavior. Gateways bridge supported private sources for managed cloud refresh workflows.
Question 720
A global catalog must balance replica freshness against latency. After selecting the proper Cosmos DB consistency choices approach, which observation gives the clearest evidence of success?
- Read tests match documented staleness and latency expectations.
- Document queries behave under partition and JSON item semantics.
- A prior version is available and current version remains identifiable.
- All three client patterns work under their expected APIs.
Explanation
Correct Answer
A:Read tests match documented staleness and latency expectations. This is the directly observable result for Cosmos DB consistency choices; cosmos DB consistency levels trade freshness availability latency and throughput.
Incorrect Answers
B:Document queries behave under partition and JSON item semantics. This supports Cosmos DB API for NoSQL versus SQL text, not confirmation of Cosmos DB consistency choices. The relevant evidence is that read tests match documented staleness and latency expectations.
C:A prior version is available and current version remains identifiable. This supports Blob versioning on overwrite, not confirmation of Cosmos DB consistency choices. The relevant evidence is that read tests match documented staleness and latency expectations.
D:All three client patterns work under their expected APIs. This supports Choosing Blob Files and Table for app needs, not confirmation of Cosmos DB consistency choices. The relevant evidence is that read tests match documented staleness and latency expectations.