Microsoft DP-700 Practice Test 37
Question 360
A semantic model refresh failed after an upstream warehouse schema change. What should be investigated first?
- Identify partial writes and replay with idempotent overwrite or merge semantics.
- Restore the referenced storage target or repair the shortcut mapping before changing notebook transforms.
- Compare model table and column dependencies with the current warehouse schema.
- Use the source event's effective business date for attribution and controlled prior-period restatement.
Explanation
Correct Answer
C:The model can fail processing when its query references changed objects even if the warehouse is otherwise healthy. Run the model's source query against the new schema.
Incorrect Answers
A:Partial destination state can persist after canceled runs; unqualified append would duplicate records.
B:A path-resolution failure happens below the transformation; notebook logic cannot read data that the shortcut no longer resolves.
D:Arrival date and business effective date differ; conflating them misstates historical periods.
Question 361
A notebook consumes a shortcut that now points to missing source data. Which layer should be repaired?
- Match pipeline activity, Spark application, and job run identifiers and timestamps.
- Restore the referenced storage target or repair the shortcut mapping before changing notebook transforms.
- Identify partial writes and replay with idempotent overwrite or merge semantics.
- Compare model table and column dependencies with the current warehouse schema.
Explanation
Correct Answer
B:A path-resolution failure happens below the transformation; notebook logic cannot read data that the shortcut no longer resolves. Test the shortcut directly outside the notebook.
Incorrect Answers
A:Cross-layer IDs join orchestration intent to actual notebook execution so failures aren't attributed to another concurrent run.
C:Partial destination state can persist after canceled runs; unqualified append would duplicate records.
D:The model can fail processing when its query references changed objects even if the warehouse is otherwise healthy.
Question 362
A late-arriving batch is mistakenly counted in today's business date. What type of defect should be corrected?
- Compare model table and column dependencies with the current warehouse schema.
- Use the source event's effective business date for attribution and controlled prior-period restatement.
- Identify partial writes and replay with idempotent overwrite or merge semantics.
- Match pipeline activity, Spark application, and job run identifiers and timestamps.
Explanation
Correct Answer
B:Arrival date and business effective date differ; conflating them misstates historical periods. Reconcile the late batch to its original business period.
Incorrect Answers
A:The model can fail processing when its query references changed objects even if the warehouse is otherwise healthy.
C:Partial destination state can persist after canceled runs; unqualified append would duplicate records.
D:Cross-layer IDs join orchestration intent to actual notebook execution so failures aren't attributed to another concurrent run.
Question 363
A team responds to intermittent throttling by increasing concurrency and sees more failures. What should it try instead?
- Identify partial writes and replay with idempotent overwrite or merge semantics.
- Match pipeline activity, Spark application, and job run identifiers and timestamps.
- Lower or bound concurrency and add backoff according to source service limits.
- Check data-quality and reconciliation results in addition to technical run status.
Explanation
Correct Answer
C:Throttled upstream systems often need regulated request rates; greater parallelism can worsen the failure. Compare throttle rate at two controlled concurrency levels.
Incorrect Answers
A:Partial destination state can persist after canceled runs; unqualified append would duplicate records.
B:Cross-layer IDs join orchestration intent to actual notebook execution so failures aren't attributed to another concurrent run.
D:A green orchestration status cannot prove business correctness when joins, late data, or missing sources alter totals.
Question 364
An on-call engineer finds conflicting evidence from a pipeline log and notebook execution. What should be correlated?
- Lower or bound concurrency and add backoff according to source service limits.
- Compare model table and column dependencies with the current warehouse schema.
- Match pipeline activity, Spark application, and job run identifiers and timestamps.
- Identify partial writes and replay with idempotent overwrite or merge semantics.
Explanation
Correct Answer
C:Cross-layer IDs join orchestration intent to actual notebook execution so failures aren't attributed to another concurrent run. Trace one reproducible failed job end to end.
Incorrect Answers
A:Throttled upstream systems often need regulated request rates; greater parallelism can worsen the failure.
B:The model can fail processing when its query references changed objects even if the warehouse is otherwise healthy.
D:Partial destination state can persist after canceled runs; unqualified append would duplicate records.
Question 365
A Delta table receives many small Parquet files after frequent incremental loads, slowing reads. Which maintenance operation fits?
- Consider Z-order optimization on the frequently filtered column.
- Retain a supported safety window that protects time travel and concurrent readers.
- Schedule supported table-maintenance work using a notebook or pipeline activity after successful loads.
- Run OPTIMIZE to compact small Delta data files.
Explanation
Correct Answer
D:Compaction reduces file-open and metadata overhead while retaining the logical table contents. Measure file count and scan time before and after.
Incorrect Answers
A:Z-order colocates values to improve file pruning for selective predicates when workload characteristics justify it.
B:Aggressive vacuuming can remove files still required for recovery or readers; default retention is typically seven days.
C:Maintenance should occur at a controlled cadence and after writes, not during arbitrary concurrent file operations.
Question 366
A lakehouse contains thousands of obsolete files no longer referenced by the Delta log. Which maintenance operation reclaims eligible storage?
- Convert or load the dataset into a supported Delta table before applying Delta maintenance.
- Run VACUUM with a safe retention period compatible with readers and time travel.
- Consider V-order when optimizing that serving Delta table.
- Consider Z-order optimization on the frequently filtered column.
Explanation
Correct Answer
B:VACUUM removes old unreferenced files, unlike OPTIMIZE, which compacts active data files. Review retention and time-travel needs before cleanup.
Incorrect Answers
A:Delta table maintenance operates on Delta transaction-log tables, not arbitrary raw file directories.
C:V-order can improve cross-engine scanning and compression for read-heavy Power BI/SQL workloads at some write cost.
D:Z-order colocates values to improve file pruning for selective predicates when workload characteristics justify it.
Question 367
A Direct Lake reporting table is read-heavy and rarely updated. Which optional table layout optimization may help?
- Retain a supported safety window that protects time travel and concurrent readers.
- Run OPTIMIZE to compact small Delta data files.
- Consider V-order when optimizing that serving Delta table.
- Run VACUUM with a safe retention period compatible with readers and time travel.
Explanation
Correct Answer
C:V-order can improve cross-engine scanning and compression for read-heavy Power BI/SQL workloads at some write cost. Benchmark reporting queries and write overhead.
Incorrect Answers
A:Aggressive vacuuming can remove files still required for recovery or readers; default retention is typically seven days.
B:Compaction reduces file-open and metadata overhead while retaining the logical table contents.
D:VACUUM removes old unreferenced files, unlike OPTIMIZE, which compacts active data files.
Question 368
A Fabric Runtime 1.2 Delta table with an established Z-order maintenance workflow is repeatedly filtered on customer_id. Which file-layout operation may improve file skipping?
- Run VACUUM with a safe retention period compatible with readers and time travel.
- Run OPTIMIZE to compact small Delta data files.
- Consider Z-order optimization on the frequently filtered column.
- Retain a supported safety window that protects time travel and concurrent readers.
Explanation
Correct Answer
C:Z-order colocates values to improve file pruning for selective predicates when workload characteristics justify it. Compare skipped files for a customer-filtered query.
Incorrect Answers
A:VACUUM removes old unreferenced files, unlike OPTIMIZE, which compacts active data files.
B:Compaction reduces file-open and metadata overhead while retaining the logical table contents.
D:Aggressive vacuuming can remove files still required for recovery or readers; default retention is typically seven days.
Question 369
A team wants to run Delta maintenance every weekend immediately after ingestion. Which orchestration pattern fits?
- Schedule supported table-maintenance work using a notebook or pipeline activity after successful loads.
- Consider V-order when optimizing that serving Delta table.
- Consider Z-order optimization on the frequently filtered column.
- Run VACUUM with a safe retention period compatible with readers and time travel.
Explanation
Correct Answer
A:Maintenance should occur at a controlled cadence and after writes, not during arbitrary concurrent file operations. Check maintenance appears in monitoring history.
Incorrect Answers
B:V-order can improve cross-engine scanning and compression for read-heavy Power BI/SQL workloads at some write cost.
C:Z-order colocates values to improve file pruning for selective predicates when workload characteristics justify it.
D:VACUUM removes old unreferenced files, unlike OPTIMIZE, which compacts active data files.
Question 370
An administrator proposes VACUUM with a few minutes of retention while consumers rely on Delta time travel. What is the safer choice?
- Consider Z-order optimization on the frequently filtered column.
- Schedule supported table-maintenance work using a notebook or pipeline activity after successful loads.
- Retain a supported safety window that protects time travel and concurrent readers.
- Consider V-order when optimizing that serving Delta table.
Explanation
Correct Answer
C:Aggressive vacuuming can remove files still required for recovery or readers; default retention is typically seven days. Inspect retention policy and time-travel requirements.
Incorrect Answers
A:Z-order colocates values to improve file pruning for selective predicates when workload characteristics justify it.
B:Maintenance should occur at a controlled cadence and after writes, not during arbitrary concurrent file operations.
D:V-order can improve cross-engine scanning and compression for read-heavy Power BI/SQL workloads at some write cost.
