DP-900 Describe an analytics workload on Azure: 302 practice questions

12 of the 302 Describe an analytics workload on Azure questions in the Certsqill DP-900 bank, shown in full below. Each one carries an explanation for every option, not just the correct one — the wrong answers are where the marks go.

Preparing for DP-900? Take the free 5-min readiness check →

1. Stream processing: What type of workload is this?

1. Batch processing

   Batch processing handles bounded data accumulated for a scheduled or defined run.
2. Stream processing ✓

   Stream processing handles an ongoing event flow and produces timely insights while events continue arriving.
3. Archival storage

   Archival storage preserves data for later access but does not analyze incoming events continuously.
4. Relational normalization

   Normalization organizes relational data structures and does not describe continuous event analysis.

This is stream processing because events arrive continuously and require timely analysis.

2. Bar or column chart: Which visual is the clearest choice?

1. Bar or column chart ✓

   Bar and column charts clearly compare values across discrete categories such as neighborhoods.
2. Line chart for time trends

   A line chart is primarily suited to trends across an ordered axis, commonly time, not this category comparison.
3. Map

   A map emphasizes geographic distribution, while the requirement is direct comparison of category values.
4. Card

   A card emphasizes one key number rather than comparing several neighborhood categories.

Use a bar or column chart to compare counts across neighborhoods.

3. Transformation: Which activity addresses the required mapping and validation?

1. Visualization

   Visualization presents prepared data through charts, rather than changing incoming shipment records.
2. Transformation ✓

   Transformation validates values and changes source structures into the analytical format required downstream.
3. Orchestration

   Orchestration coordinates tasks and dependencies, but the mapping itself is transformation work.
4. Ingestion

   Ingestion moves data from sources, but does not necessarily validate or reshape every value.

Transformation changes and validates incoming data before analytical use.

4. A near-real-time streaming workload: Which workload best fits this requirement?

1. A near-real-time streaming workload ✓

   Streaming continuously processes events, enabling timely detection without waiting for accumulated data or scheduled execution.
2. A scheduled batch job

   A scheduled batch job processes accumulated bounded data, so alerts may wait until its next execution.
3. An online transaction workload

   An online transaction workload records operational changes, rather than continuously analyzing event patterns.
4. A monthly analytical report

   A monthly report summarizes historical information and cannot support prompt operational response.

Streaming analyzes ongoing events promptly, unlike scheduled processing of accumulated data.

5. A Power BI report: Which item should the college share?

1. A dashboard

   A dashboard is a single service canvas, so it does not represent the requested multi-page analysis.
2. A data pipeline

   A data pipeline moves or transforms data, rather than providing an interactive analysis for college users.
3. A Power BI report ✓

   A report supports multiple interactive pages, visual exploration, and filtering through its associated semantic model.
4. A semantic model

   A semantic model stores analytical structure, but it is not the multi-page interactive presentation requested.

A report provides multiple interactive pages for exploration in the Power BI service.

6. A data lake: Which choice best fits this initial analytical storage need?

1. A data lake ✓

   A data lake accommodates varied raw files before analysts decide which structures and transformations are needed.
2. A Power BI dashboard

   A dashboard displays selected insights, but does not provide general-purpose storage for studio files.
3. A data warehouse

   A warehouse is optimized for modeled analytical tables, requiring more defined structure than this initial collection.
4. An OLTP database

   An OLTP database supports operational transactions, not flexible storage of diverse analytical source files.

A data lake stores diverse files flexibly before modeling and analytical preparation.

7. Azure Event Hubs: Which Azure service is the most appropriate ingestion component?

1. Azure Stream Analytics

   Stream Analytics processes and analyzes event data, but the requirement identifies the ingestion entry point.
2. A relational warehouse

   A warehouse stores modeled analytical data, rather than serving as the specified high-volume event-ingestion entry point.
3. Power BI

   Power BI visualizes analytical data, but it is not the retailer's general event-ingestion service.
4. Azure Event Hubs ✓

   Event Hubs is designed to ingest high-volume event streams before downstream processing and analytical services.

Event Hubs ingests high-volume event streams for downstream processing.

8. A Power BI dashboard: What should the clinic use?

1. A data pipeline

   A data pipeline prepares or moves data, but it does not present consolidated monitoring tiles to managers.
2. A semantic model

   A semantic model supplies data structure and calculations, but it is not the requested monitoring presentation.
3. A report page

   A report page belongs to one report experience and does not match the requested service canvas of pinned tiles.
4. A Power BI dashboard ✓

   A dashboard provides one service canvas with tiles that can be pinned from multiple reports or models.

A dashboard is a single service canvas that can combine pinned tiles from multiple sources.

9. A lakehouse: Which concept best matches this requirement?

1. A data lake

   A data lake stores varied files, but alone does not express the requested managed analytical table capabilities.
2. A warehouse

   A warehouse emphasizes relational analytical tables and does not represent the requested combined file and table experience.
3. A transactional database

   A transactional database serves operational consistency and updates, rather than combining analytical files and tables.
4. A lakehouse ✓

   A lakehouse combines flexible lake files with managed analytical tables and multiple processing access patterns.

A lakehouse combines lake-style files with managed analytical tables in one environment.

10. The events at 10: Which statement correctly describes the resulting groups?

1. All five events produce one cumulative result because streaming windows retain every event indefinitely.

   A tumbling window closes and starts another interval, so events are not placed into one indefinite cumulative group.
2. The events at 10:01 and 10:03 form one group; the remaining three form the next group. ✓

   Five-minute windows commonly cover 10:00–10:05 and 10:05–10:10, placing events into two non-overlapping groups.
3. The events at 10:03 and 10:05 must share the first group because both occur within five minutes.

   The boundary at 10:05 begins the next tumbling interval under the stated window convention.
4. Each event produces a separate result because stream processing cannot aggregate multiple events together.

   Stream processing can aggregate multiple events when a window defines the grouping interval.

A five-minute tumbling window creates separate non-overlapping intervals at each boundary.

11. Model relationship: Which design feature supplies this behavior?

1. A report page

   A report page organizes visuals and navigation; it does not establish table relationships for filter propagation.
2. Model relationship ✓

   A relationship connects the product dimension to the sales fact and allows the category filter to affect related sales.
3. A dashboard tile for display

   A dashboard tile displays results but does not define relationships or propagate filters through the semantic model.
4. A data pipeline trigger for refresh

   A pipeline trigger can start data movement or processing, but it does not define report filter propagation.

A model relationship enables the product filter to affect related sales.

12. OneLake: Which Fabric concept meets this requirement?

1. OneLake ✓

   OneLake provides unified logical lake storage across Fabric workloads, rather than acting as a relational engine.
2. A warehouse

   A warehouse provides relational analytical storage and SQL querying, which exceeds the stated shared-storage requirement.
3. A dashboard

   A dashboard presents visual tiles and does not provide shared storage for Fabric workloads.
4. A semantic model

   A semantic model defines analytical tables, relationships, and calculations, rather than unified physical lake storage.

OneLake is Fabric’s unified logical lake storage, not a relational query engine.

Other DP-900 domains

• Describe core data concepts — 302 questions →
• Identify considerations for relational data on Azure — 260 questions →
• Describe considerations for working with non-relational data on Azure — 216 questions →
• All 1080 DP-900 questions →
• DP-900 certification: requirements, cost and exam format →