Microsoft Certified: Fabric Data Engineer Associate - Full Practice Exam 1

1. A batch captures a high watermark before reading a bounded change interval. Which practices support safe progress tracking? Select 2 answers.
2. A batch upsert receives multiple source rows for the same target key. What should the engineer do before applying it?
3. A column should be obscured for ordinary SQL readers while privileged readers retain full access. Which capability should be evaluated, without treating it as complete authorization?
4. A Copy activity can partition a large source extraction. What should guide the partition key choice?
5. A Copy activity is retried after an uncertain timeout. Why should the target operation be idempotent?
6. A copy pipeline is slow, and monitoring shows the source system is saturated while the target is mostly idle. Which response is best?
7. A data engineer must create and edit workspace items but should not manage workspace membership. Which role is the usual least-privilege starting point?
8. A data product should reference an existing supported OneLake location without another ingestion copy. Which feature fits?
9. A dataflow fails after a source column changes from integer to text. Which investigation is most relevant?
10. A Delta table contains thousands of tiny files after frequent small writes. What maintenance should be evaluated?
11. A deployment pipeline promotes a data pipeline to test. Its source connection differs from development. What must be validated?
12. A dimension must retain historical customer classifications for as-of reporting. Which loading strategy fits?
13. A Fabric capacity throttles when ingestion, Spark, and reporting peak together. Which responses should be evaluated? Select 2 answers.
14. A file-arrival event should start ingestion rather than a once-daily timer. Which approach fits?
15. A KQL ingestion transform must add a normalized lowercase service name while retaining other columns. Which operator is appropriate?
16. A KQL query summarizes a year of events before applying a filter on the original event timestamp. What should be considered?
17. A large fact table joins a genuinely small dimension that fits safely in executor memory. Which optimization may avoid a costly large shuffle?
18. A nightly source is small and has no reliable change indicator. Which initial loading design is simplest if replacement is acceptable?
19. A notebook depends on Python packages that must be consistent across runs. What should the team manage explicitly?
20. A notebook produces duplicate target rows only after retries. Which issue is most likely?
21. A pipeline must copy files, execute a notebook, and refresh a downstream model in sequence. What should express the ordering?
22. A pipeline must run a notebook only after a source copy succeeds. A copy failure should alert operators instead. What should be configured?
23. A pipeline suddenly fails after a source credential rotation. Which diagnostic step is most useful first?
24. A pipeline's error handler sends a success message whenever an activity completes, even after failure. What should be corrected?
25. A PySpark input has several versions per AccountID. The latest UpdatedAt wins, with Sequence as a deterministic tie-breaker. Which design elements are needed? Select 2 answers.
26. A semantic model refresh fails after a successful pipeline load. What should monitoring distinguish?
27. A service identity executes a pipeline successfully in development but fails in production. What should be inspected first?
28. A Spark notebook repeatedly collects a large DataFrame to the driver and fails with memory errors. Which design change best addresses the cause?
29. A Spark stage has one long-running task while most tasks finish quickly. What should the engineer investigate?
30. A SQL database project is introduced for a warehouse. What is its principal lifecycle benefit?
31. A SQL ingestion query converts a text value to a number but malformed rows should be quarantined rather than abort the whole batch. Which approach fits?
32. A SQL query scans many columns and rows but the consumer needs only two columns for one date. Which change should be evaluated first?
33. A stream aggregates purchases into nonoverlapping five-minute intervals. Which window type fits?
34. A streaming backlog grows while input exceeds sustainable processing capacity. What should be examined?
35. A streaming job limits state using an event-time watermark. What tradeoff must be understood?
36. A supported operational database should be replicated into Fabric with ongoing changes rather than custom scheduled full extracts. Which capability should be evaluated?
37. A table receives repeated deliveries of the same business event. What should drive deduplication?
38. A team needs low-code routing and transformation of incoming event data into supported destinations. Which Fabric capability fits?
39. A team wants trusted data assets discoverable as reviewed organizational sources. Which feature is relevant?
40. A Warehouse query becomes inefficient after a substantial data-distribution change. What should be reviewed along with its plan?
41. A workspace is connected to Git for supported item definitions. What does this primarily provide?
42. Analysts must see only their assigned region in a supported SQL serving layer. Which control addresses this?
43. One pipeline should process different source tables using the same definition. Which design supports this?
44. Orders and Customers are joined on CustomerID. Customers contains duplicate IDs. What should the engineer check before summing revenue?
45. Several pipelines need the same ingestion routine with different parameters. What is a maintainable orchestration pattern?
46. Source records can arrive late with older event timestamps. Why may a simple maximum-event-time checkpoint miss them?
47. Spark Structured Streaming uses a durable checkpoint location. What does it support?
48. Two teams have separate workspaces on one Fabric capacity. A reporting slowdown coincides with a large Spark batch. What should the review consider?
49. Which controls improve scheduled pipeline reliability? Select 2 answers.
50. Which signals help validate an ingestion optimization? Select 2 answers.

Comments — 0