10 exam-style questions with answers and explanations, straight from our 1,030-question bank. Tap an answer to check yourself. When you're ready, take the scored version in the free practice test.
These 10 free DP-700 questions are organized by exam domain, so you can see how each part of the Microsoft Certified: Fabric Data Engineer Associate blueprint is tested. Reveal the answer and explanation under each question.
A contract analyst needs to query only reporting.vw_DailyRevenue in a Fabric warehouse. The view already exposes the approved rows and columns. The analyst has no existing access through individual or group assignments. Access to other warehouse objects and other workspace items must not be granted. How should access be provisioned?
Correct answer: A - Grant Read on the warehouse item and SELECT on reporting.vw_DailyRevenue.
Six independent table loads must run in parallel in a Fabric pipeline. ForEach iterates over objects containing a table field. Each iteration currently performs these dependent activities: Set variable: currentTable = @item().table Copy source table: @variables('currentTable') Parallel runs intermittently copy the wrong table; sequential runs succeed. What removes the race condition without serializing the loads?
Correct answer: B - Set the Copy source table directly from @item().table and remove currentTable.
An approved notebook change has been merged into the Git branch connected to the Test workspace. Fabric shows that the branch is ahead of the workspace; Test has no uncommitted changes. Test and Production are paired stages in a deployment pipeline, with their environment-specific connections already configured. Release policy permits promotion only after the new notebook version passes its tests in Test. Which sequence releases the approved change?
Correct answer: C - Update Test from Git, validate it, then deploy it to Production.
Finance maintains two analytical tables in Microsoft Fabric: CustomerBalance and AdjustmentAudit. Each adjustment must update a balance and insert its audit record. If either operation fails, neither change may be committed. The team must implement the changes in T-SQL. Which design meets these requirements?
Correct answer: C - Use a Fabric warehouse and execute both changes within one explicit T-SQL transaction.
An incremental customer load stops with the following recovery state: Durable watermark: 08:00 Retained staging batch: changes after 08:00 through 08:15, one row per non-null CustomerID Target MERGE: commit status uncertain after a connection failure Additional source changes: available through 08:30 The MERGE updates matched rows to the staged values and inserts unmatched rows. The batch is immutable, no later load has run, and a full reload cannot fit the recovery window. Select the recovery that avoids skipped changes and duplicate customer rows.
Correct answer: B - Replay the MERGE, validate the result, then advance the watermark to 08:15.
A Type 2 Customer dimension in a Fabric warehouse contains these versions for CustomerID C204: CustomerKey 81: West; ValidFrom 2026-01-01; ValidTo 2026-07-01; IsCurrent = 0 CustomerKey 144: East; ValidFrom 2026-07-01; ValidTo 9999-12-31; IsCurrent = 1 The fact input f has SaleTime = 2026-06-18 and LoadedAt = 2026-08-02. Validity intervals include their start and exclude their end. The fact table must retain one row per sale and attribute revenue to the territory at the time of sale. With d representing the dimension, select the correct join predicate.
Correct answer: A - f.CustomerID = d.CustomerID AND f.SaleTime >= d.ValidFrom AND f.SaleTime < d.ValidTo
Calculate the percentage of individual requests that failed for each service, using an Eventhouse table named ApiIntervals. Its columns are Service:string, Requests:long, and Failures:long. Each row holds counts for a different, nonoverlapping interval; every request appears in exactly one interval. Requests is positive, and Failures counts failed requests. Traffic volume and failure rates vary between intervals. Which KQL query implements the required service-level metric?
Correct answer: C - ApiIntervals | summarize FailurePct = 100.0 * sum(Failures) / sum(Requests) by Service
The Spark UI for a Fabric notebook shows this pattern during a sort-merge join on CustomerID: 199 join tasks: finish within 20 seconds 1 join task: runs for 18 minutes and spills heavily Rows with CustomerID = 0: 58% of the left input, sent to the slow task Capacity throttling: none Adaptive Query Execution: disabled Both join inputs are too large to broadcast within the available executor memory. What change most directly addresses the bottleneck?
Correct answer: D - Enable Adaptive Query Execution and its skew-join optimization.
Dataflow Gen2 converts a supplier's day/month/year text dates using the English (United States) locale. The results include: 31/03/2026 -> conversion error 03/04/2026 -> March 4, although the supplier intended April 3 The connection is healthy, and retries produce identical results. The supplier cannot reissue this delivery before today's processing deadline. Which repair addresses both the visible error and the silently incorrect date?
Correct answer: D - Convert the original text to Date using the English (United Kingdom) locale.
An audit query began reading a Delta lakehouse table at 09:00 and must finish against that recorded snapshot. Subsequent writes changed the current data, and an 11:00 OPTIMIZE made some files needed by the audit query obsolete in the latest version. At 12:00, the query is still running and a cleanup job is about to execute VACUUM RETAIN 0 HOURS with its retention safety check disabled. Storage capacity is sufficient to retain the old files. What should the on-call engineer do?
Correct answer: D - Defer VACUUM; retain the files required by the running audit query.
The full bank has 1,020 more DP-700 questions with explanations.
Continue in the free practice test →
View plans