cert
cert-1
cert-2

Pass Microsoft DP-900 Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers to Pass the Actual Exam!
30 Days Free Updates, Instant Download!

cert-5
cert-6
DP-900 Exam - Verified By Experts
DP-900 Premium Bundle
$29.98

DP-900 Premium Bundle

$79.99
$109.97
  • Premium File 332 Questions & Answers. Last update: Nov 13, 2026
  • Training Course 32 Video Lectures
  • Study Guide 672 Pages
 
$109.97
$79.99
accept 21 downloads in last 7 days
block-screenshots
Purchase Microsoft DP-900 Exam Training Products Individually
Microsoft DP-900 Premium File
Premium File 332 Q&A
$76.99$69.99
Microsoft DP-900 Video Training Course
Training Course 32 Lectures
$21.99 $19.99
Microsoft DP-900 PDF Study Guide
Study Guide 672 Pages
$21.99 $19.99
Microsoft DP-900 Practice Test 90

Question 890

Incoming equipment events must be filtered and aggregated every minute. Which choice BEST meets this requirement?

  1. Use a pie or donut only when meaningful proportions have few categories.
  2. Rename an Azure SQL Database query Stream and assume it automatically subscribes to Event Hubs.
  3. Use Azure Stream Analytics job inputs queries and compatible outputs.
  4. Use Azure Event Hubs for eligible high-throughput event ingestion.

Explanation

Correct Answer

C:

Use Azure Stream Analytics job inputs queries and compatible outputs. Stream Analytics analyzes streams using continuous SQL-like queries. In the stated case, verification should show that output windows contain correct event aggregates under steady ingestion.

Incorrect Answers

A:

Use a pie or donut only when meaningful proportions have few categories. That action primarily handles Pie donut chart proportion caution; the given requirement is Azure Stream Analytics SQL-like streaming. Stream Analytics analyzes streams using continuous SQL-like queries.

B:

Rename an Azure SQL Database query Stream and assume it automatically subscribes to Event Hubs. It does not implement Azure Stream Analytics SQL-like streaming. A single historical SELECT does not continuously process future events.

D:

Use Azure Event Hubs for eligible high-throughput event ingestion. That action primarily handles Azure Event Hubs event ingestion; the given requirement is Azure Stream Analytics SQL-like streaming. Stream Analytics analyzes streams using continuous SQL-like queries.

Question 891

A pagination job returns duplicate or missing records because ties have no ordering. Which principle BEST explains the correct SQL ORDER BY pagination stability design decision?

  1. SQL row order is not guaranteed without ordering clauses.
  2. SQL NULL represents absent or unknown values and ordinary equality does not match it.
  3. Managed Instance has deployment/networking characteristics that differ from a single SQL Database.
  4. Azure SQL Database abstracts most server and engine patch operations.

Explanation

Correct Answer

A:

SQL row order is not guaranteed without ordering clauses. In this case the applicable decision is to specify ORDER BY with a deterministic tie-breaker when paging.

Incorrect Answers

B:

SQL NULL represents absent or unknown values and ordinary equality does not match it. That concept explains NULL and three-valued logic; the situation instead concerns SQL ORDER BY pagination stability. Unstable ordering across ties can skip or repeat rows.

C:

Managed Instance has deployment/networking characteristics that differ from a single SQL Database. That concept explains SQL Managed Instance virtual network; the situation instead concerns SQL ORDER BY pagination stability. Unstable ordering across ties can skip or repeat rows.

D:

Azure SQL Database abstracts most server and engine patch operations. That concept explains Azure SQL Database single database PaaS; the situation instead concerns SQL ORDER BY pagination stability. Unstable ordering across ties can skip or repeat rows.

Question 892

A transaction fails after creating the order header but before allocating stock. What is the MOST appropriate next validation or correction?

  1. Test document fetches separately from invoice property filtering.
  2. Treat a Power BI refresh as the commit step for a credit card purchase.
  3. Collect latency distributions and batch capacity under realistic load.
  4. Test rollback when the second operation fails.

Explanation

Correct Answer

D:

Test rollback when the second operation fails. This tests the intended result: both changes commit or neither change remains partially recorded.

Incorrect Answers

A:

Test document fetches separately from invoice property filtering. It tests File store versus database selection rather than the critical OLTP atomic purchase transaction behavior. Transactional workloads emphasize consistent low-latency create update delete operations.

B:

Treat a Power BI refresh as the commit step for a credit card purchase. This unsafe shortcut leaves the OLTP atomic purchase transaction problem unresolved. Partial independent commits can oversell inventory.

C:

Collect latency distributions and batch capacity under realistic load. It tests Operational SLA latency versus throughput rather than the critical OLTP atomic purchase transaction behavior. Transactional workloads emphasize consistent low-latency create update delete operations.

Question 893

An analyst queries eventhouse log streams to find rising error rates. What is the MOST appropriate next validation or correction?

  1. Trace sample errors through each refinement layer.
  2. Assume changing the color of a dashboard chart performs KQL aggregation.
  3. Test producer ingestion and consumer offsets across partitions.
  4. Compare summarize results to controlled event sets.

Explanation

Correct Answer

D:

Compare summarize results to controlled event sets. This tests the intended result: counts match known events and time buckets.

Incorrect Answers

A:

Trace sample errors through each refinement layer. It tests Bronze silver gold medallion design rather than the critical KQL summarize time-series events behavior. KQL enables time-oriented event aggregation for operational analytics.

B:

Assume changing the color of a dashboard chart performs KQL aggregation. This unsafe shortcut leaves the KQL summarize time-series events problem unresolved. Counting events without time filtering can misstate incident scope.

C:

Test producer ingestion and consumer offsets across partitions. It tests Azure Event Hubs event ingestion rather than the critical KQL summarize time-series events behavior. KQL enables time-oriented event aggregation for operational analytics.

Question 894

A published report still shows last week's sales due to a failed refresh. Which principle BEST explains the correct Power BI scheduled refresh and gateway design decision?

  1. ELT loads before transformation and uses destination processing resources.
  2. Line charts effectively show change across time.
  3. Gateways bridge supported private sources for managed cloud refresh workflows.
  4. Event time differs from ingestion or compute processing time.

Explanation

Correct Answer

C:

Gateways bridge supported private sources for managed cloud refresh workflows. In this case the applicable decision is to configure an appropriate data gateway and scheduled refresh where needed.

Incorrect Answers

A:

ELT loads before transformation and uses destination processing resources. That concept explains ELT warehouse pushdown transforms; the situation instead concerns Power BI scheduled refresh and gateway. Publishing a report does not guarantee the local source is reachable.

B:

Line charts effectively show change across time. That concept explains Line chart time-series selection; the situation instead concerns Power BI scheduled refresh and gateway. Publishing a report does not guarantee the local source is reachable.

D:

Event time differs from ingestion or compute processing time. That concept explains Event time versus processing time; the situation instead concerns Power BI scheduled refresh and gateway. Publishing a report does not guarantee the local source is reachable.

Question 895

An IT team wants to recover a previous version of a shared directory file. During testing, the team observes this specific failure: Only replicating the latest share state may not recover accidental overwrites. Which corrective check should be performed FIRST?

  1. Use Blob versioning where supported and protect earlier object versions.
  2. Name a folder Snapshot and assume Azure Files retains its prior contents.
  3. Perform controlled snapshot restore test for a sample share.
  4. Evaluate Cosmos DB for Table with supported API compatibility and cost tradeoffs.

Explanation

Correct Answer

C:

Perform controlled snapshot restore test for a sample share. A successful correction should demonstrate that a selected older file state can be restored within policy.

Incorrect Answers

A:

Use Blob versioning where supported and protect earlier object versions. It serves Blob versioning on overwrite, not the diagnosed Azure Files snapshots and backups problem. Only replicating the latest share state may not recover accidental overwrites.

B:

Name a folder Snapshot and assume Azure Files retains its prior contents. This reinforces the error rather than addressing the Azure Files snapshots and backups failure. File share snapshots preserve point-in-time share states when configured.

D:

Evaluate Cosmos DB for Table with supported API compatibility and cost tradeoffs. It serves Cosmos DB API for Table compatibility, not the diagnosed Azure Files snapshots and backups problem. Only replicating the latest share state may not recover accidental overwrites.

Question 896

A collection frequently filters by nested order status. After selecting the proper Cosmos DB NoSQL indexing and queries approach, which observation gives the clearest evidence of success?

  1. New entries extend the log without rewriting an unrelated object.
  2. Graph queries follow intended relationship paths and return connected entities.
  3. Recovery expectations match documented replica placement and failover support.
  4. Intended queries work with measurable RU use under representative data.

Explanation

Correct Answer

D:

Intended queries work with measurable RU use under representative data. This is the directly observable result for Cosmos DB NoSQL indexing and queries; indexed document properties support queries while policy affects write/read cost.

Incorrect Answers

A:

New entries extend the log without rewriting an unrelated object. This supports Append blob sequential logging, not confirmation of Cosmos DB NoSQL indexing and queries. The relevant evidence is that intended queries work with measurable RU use under representative data.

B:

Graph queries follow intended relationship paths and return connected entities. This supports Cosmos DB API for Apache Gremlin graph, not confirmation of Cosmos DB NoSQL indexing and queries. The relevant evidence is that intended queries work with measurable RU use under representative data.

C:

Recovery expectations match documented replica placement and failover support. This supports Azure Storage LRS ZRS GRS distinctions, not confirmation of Cosmos DB NoSQL indexing and queries. The relevant evidence is that intended queries work with measurable RU use under representative data.

Question 897

A service needs a direct lookup by device key without SQL joins. Select TWO statements that correctly describe the Table Storage PartitionKey and RowKey implementation and how it can be verified.

  1. Consider append blobs for append-based writes with appropriate access and retention controls.
  2. Evaluate multi-region writes with a deliberate conflict resolution policy.
  3. Design entities with appropriate PartitionKey and RowKey values.
  4. Query only a free-text Description field and assume it is the unique Table primary key.
  5. Point lookups retrieve the intended entity efficiently under correct keys.

Explanation

Correct Answers

C:

Design entities with appropriate PartitionKey and RowKey values. This is an appropriate action for the Table Storage PartitionKey and RowKey case. Table Storage identifies entities by partition and row keys.

E:

Point lookups retrieve the intended entity efficiently under correct keys. This is directly observable success evidence for Table Storage PartitionKey and RowKey.

Incorrect Answers

A:

Consider append blobs for append-based writes with appropriate access and retention controls. This implements Append blob sequential logging instead of the required Table Storage PartitionKey and RowKey function. Table Storage identifies entities by partition and row keys.

B:

Evaluate multi-region writes with a deliberate conflict resolution policy. This implements Cosmos DB multi-region writes conflict handling instead of the required Table Storage PartitionKey and RowKey function. Table Storage identifies entities by partition and row keys.

D:

Query only a free-text Description field and assume it is the unique Table primary key. This is not a safe or appropriate Table Storage PartitionKey and RowKey implementation. Unbalanced partition keys or unselective scans can harm performance.

Question 898

A Spark workload performs directory rename and partition discovery operations. What is the MOST appropriate next validation or correction?

  1. Verify supported alignment and read-write range operations.
  2. Rename a container adls and assume hierarchical namespace is enabled.
  3. Confirm account namespace setting and test folder access.
  4. Test multi-edge routes and direction filters on sample graph.

Explanation

Correct Answer

C:

Confirm account namespace setting and test folder access. This tests the intended result: folder-level operations and permissions work as designed.

Incorrect Answers

A:

Verify supported alignment and read-write range operations. It tests Page blob random access rather than the critical ADLS Gen2 hierarchical namespace behavior. ADLS Gen2 layers hierarchical file namespace capabilities onto Blob storage.

B:

Rename a container adls and assume hierarchical namespace is enabled. This unsafe shortcut leaves the ADLS Gen2 hierarchical namespace problem unresolved. Treating virtual folder name prefixes as identical to hierarchical directory ACLs can fail.

D:

Test multi-edge routes and direction filters on sample graph. It tests Cosmos DB API for Apache Gremlin graph rather than the critical ADLS Gen2 hierarchical namespace behavior. ADLS Gen2 layers hierarchical file namespace capabilities onto Blob storage.

Question 899

A design team incorrectly treats event broker retention as a permanent warehouse. What is the MOST appropriate next validation or correction?

  1. Kill a test consumer and reconcile events after restart.
  2. Treat a 24-hour Event Hubs retention policy as a ten-year audit archive.
  3. Check workspace and OneLake access from representative users.
  4. Recheck events after the broker expiry and inspect long-term tables.

Explanation

Correct Answer

D:

Recheck events after the broker expiry and inspect long-term tables. This tests the intended result: historical queries remain possible after the broker retention period expires.

Incorrect Answers

A:

Kill a test consumer and reconcile events after restart. It tests Streaming checkpoints and replay rather than the critical Stream retention and historical analytics behavior. Stream brokers and analytical stores have different persistence and replay roles.

B:

Treat a 24-hour Event Hubs retention policy as a ten-year audit archive. This unsafe shortcut leaves the Stream retention and historical analytics problem unresolved. Failing to archive selected events loses history after retention.

C:

Check workspace and OneLake access from representative users. It tests Microsoft Fabric OneLake unified storage rather than the critical Stream retention and historical analytics behavior. Stream brokers and analytical stores have different persistence and replay roles.

Question 900

Developers want managed Postgres backups and version updates in Azure. What is the MOST appropriate next validation or correction?

  1. Run an isolated restore and reconcile target time and transactions.
  2. Rename a SQL Server connection string postgres and assume the protocol changes.
  3. Run representative extension and client compatibility tests.
  4. Compare plans and update costs before and after an index.

Explanation

Correct Answer

C:

Run representative extension and client compatibility tests. This tests the intended result: postgreSQL clients and queries work while managed service maintenance is planned.

Incorrect Answers

A:

Run an isolated restore and reconcile target time and transactions. It tests Azure SQL Database backup PITR rather than the critical Azure Database for PostgreSQL Flexible Server behavior. PostgreSQL services preserve engine-specific language and compatibility semantics.

B:

Rename a SQL Server connection string postgres and assume the protocol changes. This unsafe shortcut leaves the Azure Database for PostgreSQL Flexible Server problem unresolved. Treating PostgreSQL as SQL Server may break functions and drivers.

D:

Compare plans and update costs before and after an index. It tests B tree index for lookups rather than the critical Azure Database for PostgreSQL Flexible Server behavior. PostgreSQL services preserve engine-specific language and compatibility semantics.

Microsoft Azure Data Fundamentals (DP-900) Practice Tests