DP-700 - Exam-Style Scenarios¶

12 scenarios covering the major exam domains. Read the prompt, pick the best answer, then check the explanation.

Scenario 1¶

You need to continuously replicate Azure Cosmos DB data into Fabric for analytics with the lowest engineering effort. What do you configure?

Best answer: Mirroring - configure mirroring on the Cosmos DB account from the Fabric workspace. Initial backfill + ongoing CDC, no pipeline required. Lands as Delta in OneLake.

Scenario 2¶

A real-time IoT stream from Azure IoT Hub must land in a Lakehouse and also feed a real-time dashboard with sub-second latency.

Best answer: Eventstream with two destinations: a Lakehouse (for batch/long-term analytics) and a KQL Database / Eventhouse (for the sub-second dashboard via real-time queries).

Scenario 3¶

Power BI users need to query a 100GB Lakehouse table with import-like speed without the import refresh cycle.

Best answer: Use Direct Lake storage mode in Power BI. Optimize with V-Order on the Delta table.

Scenario 4¶

A data engineer needs full T-SQL DDL/DML on a table (CREATE / UPDATE / DELETE), but the table needs to live in OneLake.

Best answer: Use a Warehouse. Lakehouse SQL endpoint is read-only. Both store data as Delta in OneLake.

Scenario 5¶

You need to promote a Pipeline from Dev to Test to Prod, swapping the source data source for each environment.

Best answer: Deployment Pipelines with deployment rules that swap source parameters per environment.

Scenario 6¶

A Spark notebook job is taking 6 hours and one task is much slower than others. Diagnostic approach?

Best answer: Open the Spark UI from the notebook execution. Look at stage and task durations. Slow single task = data skew. Mitigate by salting the join key, repartitioning, or broadcasting the smaller side.

Scenario 7¶

A Cosmos DB query analyst needs read access to a specific Lakehouse, but should not see other items in the same workspace.

Best answer: Don't grant workspace role. Instead, grant Read at the item level on the specific Lakehouse only.

Scenario 8¶

Daily revenue must trigger an email alert when it drops more than 20% from yesterday.

Best answer: Reflex item watching a KQL query (or a Power BI semantic model measure) with a condition rule. Trigger email or Power Automate flow on alert.

Scenario 9¶

You want a low-code transform that filters, joins, and reshapes data from 3 SaaS sources into a Warehouse table.

Best answer: Dataflow Gen2 with output to the Warehouse. Power Query handles the filter/join/reshape; the Warehouse output destination loads the result.

Scenario 10¶

The team wants to back the workspace's content with version control and review changes via PRs.

Best answer: Connect the workspace to an Azure DevOps or GitHub repo using Fabric Git integration. Items sync as JSON/source. Commits and PRs work as expected.

Scenario 11¶

A Lakehouse table grows to 5 TB and Direct Lake queries slow down. What do you do?

Best answer: Run OPTIMIZE table VORDER on the Delta table. Optionally OPTIMIZE table ZORDER BY (col) on commonly-filtered columns. Partition the table by date if appropriate. Vacuum old versions.

Scenario 12¶

Which item type is appropriate for sub-second analytics on streaming data with Kusto query language?

Best answer: KQL Database / Eventhouse. Backed by Kusto engine. Optimized for real-time analytics on append-only event data.

Scoring guide¶

10-12: Schedule the exam.
7-9: Re-read fact-sheet and weak-area notes.
<7: More hands-on practice with Fabric Free Trial.

DP-700 has multiple-choice + multi-response + case studies + sometimes labs. These scenarios test pattern recognition; build hands-on fluency with the Fabric UI for the labs.