DP-700: Implementing Data Engineering Solutions Using Microsoft Fabric

DP-700 exam objectives and study guide

The skills measured on the DP-700 exam, by objective domain. Percentages are the share of the exam each domain carries.

DP-700 practice questions and answers

10 free sample questions from the DP-700 bank, with the correct answer and a full explanation for each. These are original questions written to the Microsoft objective domains — not real exam content.

1. Question 1Ingest and transform data

   You are preparing a Fabric notebook for semi-structured order data that will be curated with Spark and then read by analysts through the Lakehouse SQL analytics endpoint. The team selected a Lakehouse because the raw files and managed tables need to remain in OneLake. Complete the PySpark write operation so the result is a Lakehouse table rather than only a folder of files.

   (orders_df.write
       .format('1')
       .mode('append')
       .2('curated_orders'))

   Blank 1

   • Adelta
   • Bparquet
   • Cjson
   • Dcsv

   Blank 2

   • AsaveAsTable
   • Bsave
   • CinsertInto
   • Dparquet

   Show answer and explanation

   Correct answers

   • Blank 1: delta
   • Blank 2: saveAsTable

   Explanation

   delta and saveAsTable are the correct choices. Fabric Lakehouse tables are Delta Lake tables in OneLake, and saveAsTable registers the DataFrame output as a table that can be discovered and queried through table-oriented experiences such as Spark and the SQL analytics endpoint.

   • parquet, json, and csv are valid Spark formats for files, but they do not create a Delta Lake table with a transaction log.
   • save writes to a path and does not register a managed table by name.
   • insertInto appends to an existing table and is not the right operation when the notebook must create the Lakehouse table.
   • parquet as a writer method writes Parquet files to a path rather than creating the named Lakehouse table.

   References:

   • What is a lakehouse?
   • What is the SQL analytics endpoint for a lakehouse?
2. Question 2Ingest and transform data

   Marisol, an analytics engineer, must land monthly vendor packages that contain CSV manifests, PDF invoices, and nested JSON detail files. Data scientists will explore the files with Spark notebooks, and analysts only need read-only SQL over curated Delta tables. Which Fabric store should Marisol create?

   • AFabric Lakehouse
   • BFabric Warehouse
   • CKQL database in a Fabric Eventhouse
   • DMirrored Azure SQL database
   • EPower BI semantic model

   Show answer and explanation

   Correct answer

   • Fabric Lakehouse

   Explanation

   A Fabric Lakehouse is the best fit. A Lakehouse supports files and Delta tables in OneLake, Spark notebooks for engineering and exploration, and a SQL analytics endpoint for read-only SQL access to tables.

   • A Warehouse is suited to relational T-SQL analytics, but it is not the right store for keeping PDFs and raw nested files next to Spark-managed tables.
   • An Eventhouse is optimized for real-time event analytics and KQL, not monthly mixed file packages.
   • A Mirrored Azure SQL database is for replicating supported operational databases, not landing vendor file packages.
   • A semantic model is a BI modeling layer, not the physical data store for raw invoices and curated tables.

   References:

   • What is a lakehouse?
   • What is the SQL analytics endpoint for a lakehouse?
   • What Is Fabric Data Warehouse?
3. Question 3Monitor and optimize an analytics solution

   Anika, a Fabric operations lead, must create a daily operations process for semantic model refreshes across eight workspaces. Analysts need an interactive view for failed and running refresh activities in the Fabric portal, and an automation runbook must store the last refresh status for each model as JSON. Which TWO monitoring approaches should Anika include?

   • AInstall the Microsoft Fabric Capacity Metrics app and use CU utilization rows as the refresh success log.
   • BUse the Fabric Monitoring hub, filtered to semantic model refresh activities, for the operator view.
   • CSend Azure Data Factory pipeline diagnostic logs to Log Analytics and query pipeline run failures.
   • DCall the Power BI REST API Get Refresh History In Group for each semantic model and persist the returned records.
   • EScan the workspaces with Microsoft Purview Data Map and read certification metadata.

   Show answer and explanation

   Correct answers

   • Use the Fabric Monitoring hub, filtered to semantic model refresh activities, for the operator view.
   • Call the Power BI REST API Get Refresh History In Group for each semantic model and persist the returned records.

   Explanation

   The correct answers are B and D.

   • B is correct because the Fabric Monitoring hub provides a central portal experience for monitoring Fabric item activities, including semantic model refresh activity visible to users with access.
   • D is correct because the Power BI REST API Get Refresh History In Group operation returns semantic model refresh history records as JSON for models in a workspace. The REST API still uses the datasets path for semantic models.
   • A is incorrect because the Microsoft Fabric Capacity Metrics app is used to analyze capacity usage, throttling, and CU consumption. It is not the operational refresh success and error-history log for each semantic model.
   • C is incorrect because Azure Data Factory diagnostic logs cover Data Factory pipeline activity, not Power BI/Fabric semantic model refresh status.
   • E is incorrect because Microsoft Purview Data Map is for data governance, cataloging, scanning, and lineage scenarios, not runtime refresh monitoring.

   References:

   • Datasets
   • What is the Microsoft Fabric Capacity Metrics app?
4. Question 4Monitor and optimize an analytics solution

   Users report that the sales report still shows yesterday's data even though the scheduled import semantic model refresh should run at 06:00. The model uses an on-premises data gateway, and the administrator must confirm whether the scheduled refresh failed because of credentials without checking gateway machine logs. What should the administrator open first?

   • AThe Manage connections and gateways page for the gateway cluster.
   • BThe report usage metrics report for the affected report.
   • CThe semantic model Refresh history page in the Power BI service.
   • DThe workspace lineage view for the report and semantic model.

   Show answer and explanation

   Correct answer

   • The semantic model Refresh history page in the Power BI service.

   Explanation

   The correct answer is C.

   • C is correct because the semantic model Refresh history shows scheduled and on-demand refresh attempts, including status, timing, and failure details for a specific refresh run. A gateway credential failure is surfaced as a refresh failure for the model.
   • A is incorrect because Manage connections and gateways is used to administer gateway connections and credentials, but it does not provide the run-by-run refresh history for the 06:00 semantic model refresh.
   • B is incorrect because report usage metrics show report consumption behavior, not whether a semantic model refresh succeeded or failed.
   • D is incorrect because lineage view shows item dependencies. It does not provide the detailed failure message for a scheduled semantic model refresh.

   References:

   • Data refresh in Power BI
   • Configure scheduled refresh
5. Question 5Monitor and optimize an analytics solution

   Wide World Importers has a Fabric F64 capacity where two import semantic models refresh during business hours. Refresh history shows completion times are increasing, but the team needs to determine whether refresh workloads are causing CU saturation and throttling on the capacity during the same interval. Which monitoring source should they use?

   • ASemantic model Refresh history for each affected model.
   • BPower BI Desktop Performance Analyzer for the reports.
   • COneLake catalog for the workspace items.
   • DMicrosoft Fabric Capacity Metrics app for the F64 capacity.

   Show answer and explanation

   Correct answer

   • Microsoft Fabric Capacity Metrics app for the F64 capacity.

   Explanation

   The correct answer is D.

   • D is correct because the Microsoft Fabric Capacity Metrics app is designed to analyze Fabric capacity consumption, including CU usage and throttling. It is the right source for correlating semantic model refresh workloads with capacity saturation.
   • A is incorrect because semantic model Refresh history is useful for per-model status, duration, and failure details, but it does not provide the capacity-level CU saturation and throttling analysis required here.
   • B is incorrect because Power BI Desktop Performance Analyzer measures report visual and query performance during report authoring, not capacity-level refresh workload consumption.
   • C is incorrect because OneLake catalog helps users discover and manage Fabric data items. It is not a capacity utilization or throttling monitoring tool.

   References:

   • What is the Microsoft Fabric Capacity Metrics app?
   • Data refresh in Power BI
6. Question 6Monitor and optimize an analytics solution

   To verify yesterday's scheduled refresh for one semantic model, you want to use MicrosoftPowerBIMgmt in PowerShell instead of opening the portal. Complete the command that retrieves the latest refresh history records for the model.

   Connect-PowerBIServiceAccount
   $uri = "groups/$workspaceId/datasets/$semanticModelId/refreshes?`$top=5"
   Invoke-PowerBIRestMethod -Url $uri -Method 1

   Blank 1

   • APost
   • BPatch
   • CGet
   • DDelete

   Show answer and explanation

   Correct answer

   • Blank 1: Get

   Explanation

   The correct completion is Get.

   Invoke-PowerBIRestMethod can call Power BI REST API endpoints. A GET request to the refreshes collection retrieves refresh history for the semantic model. Post is used on this collection to start a refresh, not to read history. Patch and Delete do not retrieve semantic model refresh history.

   References:

   • Invoke-PowerBIRestMethod (MicrosoftPowerBIMgmt.Profile)
   • Connect-PowerBIServiceAccount (MicrosoftPowerBIMgmt.Profile)
   • Datasets
7. Question 7Monitor and optimize an analytics solution

   Fourth Coffee's finance team asks whether yesterday evening's scheduled import refresh for the Finance semantic model succeeded. Which location in the Power BI service provides this information for that model?

   • AThe report Subscriptions pane.
   • BThe semantic model Refresh history page.
   • CThe workspace Access page.
   • DThe deployment pipeline rules page.

   Show answer and explanation

   Correct answer

   • The semantic model Refresh history page.

   Explanation

   The correct answer is B.

   • B is correct because the semantic model Refresh history is the built-in place to view recent refresh attempts and their status for a specific model.
   • A is incorrect because report subscriptions control email delivery of report snapshots or links, not semantic model refresh run status.
   • C is incorrect because the workspace Access page manages permissions and does not show refresh outcomes.
   • D is incorrect because deployment pipeline rules control item deployment behavior between stages, not the refresh history of a semantic model.

   References:

   • Data refresh in Power BI
   • Configure scheduled refresh
8. Question 8Monitor and optimize an analytics solution

   You are tuning a Fabric notebook that creates three silver tables from the same cleansed bronze DataFrame. The Spark UI shows the same scan, filter, and projection stages repeated for a `count()` check and two write actions. The capacity has memory headroom, and the team does not want to persist an extra intermediate Delta table in OneLake. Which change best addresses the issue?

   • ACall `persist()` or `cache()` on the cleansed DataFrame before the first action, and call `unpersist()` after the final action.
   • BCheckpoint the cleansed DataFrame to a new Delta table and read it back for each silver-table write.
   • CIncrease `spark.sql.shuffle.partitions` so the repeated transformations run with more tasks.
   • DUse `broadcast()` on the cleansed bronze DataFrame before deriving the three silver tables.

   Show answer and explanation

   Correct answer

   • Call `persist()` or `cache()` on the cleansed DataFrame before the first action, and call `unpersist()` after the final action.

   Explanation

   Cache or persist the reused DataFrame before the first action, and unpersist it after the last consumer. Spark DataFrames are lazily evaluated, so multiple actions can recompute the same lineage unless the intermediate result is materialized in memory or storage.

   • Checkpointing or writing an intermediate Delta table can break lineage, but it violates the requirement to avoid persisting another intermediate table in OneLake.
   • Increasing spark.sql.shuffle.partitions changes shuffle parallelism; it does not prevent repeated scans of the same lineage.
   • Broadcasting the cleansed bronze DataFrame is intended for small join inputs and is not appropriate for a large shared transformation.

   References:

   • Apache Spark monitoring overview
   • What is the medallion lakehouse architecture?
9. Question 9Monitor and optimize an analytics solution

   Khalil, a warehouse performance analyst, must reduce the runtime of a Fabric Data Warehouse report after a large nightly load. Query insights shows that the same T-SQL statement is now slower, and the query filters and joins on columns whose value distribution changed during the load. The team must keep the current Fabric capacity and avoid changing the report query. What should Khalil do first?

   • ARun `UPDATE STATISTICS` on the affected Warehouse tables or statistics for the columns used by the filters and joins.
   • BRun Lakehouse table maintenance with `OPTIMIZE` and V-Order on the Warehouse tables before the report starts.
   • CInstrument the report client by using Application Insights OpenTelemetry and review Live metrics during the report run.
   • DChange the semantic model storage mode to Import so that the report stops issuing the Warehouse query.

   Show answer and explanation

   Correct answer

   • Run `UPDATE STATISTICS` on the affected Warehouse tables or statistics for the columns used by the filters and joins.

   Explanation

   The correct answer is to update the relevant Warehouse statistics. Fabric Data Warehouse query optimization depends on statistics for cardinality estimates. After a large data change, stale or missing statistics on filter and join columns can cause inefficient plans, so using T-SQL statistics statements such as UPDATE STATISTICS or creating needed statistics is the right first optimization step.

   • Running Lakehouse table maintenance with V-Order is useful for Delta tables in a Lakehouse, but it does not address stale query optimizer statistics in a Fabric Data Warehouse.
   • Adding Application Insights OpenTelemetry instrumentation can improve application observability, but it does not change the Warehouse optimizer plan or update data distribution metadata.
   • Moving the report to Import mode can sometimes reduce report-time query latency, but it violates the requirement to avoid changing the report query path and does not fix the Warehouse query regression.

   References:

   • Statistics
   • Query Insights
   • Performance Guidelines
10. Question 10Monitor and optimize an analytics solution

    Users report that a Direct Lake report becomes slower after each hourly append to a large Fabric Lakehouse table. The Spark write job creates many small Delta files, and the same table is queried by the SQL analytics endpoint. The solution must preserve Direct Lake behavior and must not increase capacity. Which maintenance action should you schedule?

    • ARun Lakehouse table maintenance with `OPTIMIZE` and V-Order for the frequently queried Delta table after the append workload.
    • BRun `VACUUM` for the table after every hourly append so that obsolete Delta files are removed immediately.
    • CRewrite the table by partitioning it on the unique session identifier used by the clickstream source.
    • DConvert the semantic model to Import mode and schedule refreshes to match the hourly append process.

    Show answer and explanation

    Correct answer

    • Run Lakehouse table maintenance with `OPTIMIZE` and V-Order for the frequently queried Delta table after the append workload.

    Explanation

    The correct answer is to schedule Lakehouse table maintenance that runs OPTIMIZE with V-Order. OPTIMIZE compacts many small Delta files into larger files, and V-Order improves read performance for Fabric engines that consume Delta/Parquet data, including Direct Lake and the SQL analytics endpoint.

    • VACUUM removes old files that are no longer needed by Delta retention, but it does not compact active small files.
    • Partitioning by a high-cardinality session identifier typically creates many small partitions and files, which worsens the symptom.
    • Switching to Import mode can reduce interactive query load, but it violates the requirement to preserve Direct Lake behavior and introduces refresh management.
    • Application Insights Live metrics are for application performance monitoring and do not optimize Delta file layout in OneLake.

    References:

    • Delta Table Maintenance in Microsoft Fabric
    • Optimize Delta Lake tables with V-Order in Fabric
    • Lakehouse and Delta Tables

Membership includes 180 questions and explanations aligned to the DP-700 curriculum, including 8 case studies.

Other Microsoft certifications

Every one of these is included with the same membership as DP-700.

AB-900Microsoft

Microsoft 365 Copilot and Agent Administration Fundamentals

This exam is intended for candidates who are familiar with Microsoft 365, including core services, security, identity and access, data…

161 practice questionsView

AI-900Microsoft

Microsoft Azure AI Fundamentals

This exam measures knowledge of machine learning and AI concepts and related Microsoft Azure services. It is intended for candidates…

193 practice questionsView

AZ-900Microsoft

Microsoft Azure Fundamentals

This exam is intended for technology professionals who want to demonstrate foundational knowledge of cloud concepts and Microsoft Azure.…

170 practice questionsView

DP-700 exam FAQ