Exam Bundle

This signal belongs to the adjacent case “A join's keys are heavily skewed to one customer ID,” whose remedy is Rebalance skew using supported AQE/skew mitigation or targeted salting. The observed result for the present requirement is The join completes without broadcast OOM and returns correct cardinality.

That observation is a valid indicator for “The join creates a Cartesian product because the ON predicate was omitted” when using Correct the equality/range join condition for the business keys. The scenario instead tests “Two large datasets need a join and memory pressure makes forced broadcast impractical”; the relevant result is The join completes without broadcast OOM and returns correct cardinality.

That observation is a valid indicator for “A query shuffles a huge wide DataFrame before filtering unneeded rows” when using Push semantically safe filters and projections ahead of the shuffle. The scenario instead tests “Two large datasets need a join and memory pressure makes forced broadcast impractical”; the relevant result is The join completes without broadcast OOM and returns correct cardinality.

The alternative Allow AQE to adapt supported join strategy at runtime has a distinct purpose—The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering. It would not establish the required outcome (Shuffle partitions are badly sized at runtime due to variable daily input); the necessary verification concerns Runtime plan coalesces undersized shuffle partitions.

This action solves a different issue: “Executors are overwhelmed by excessive task count from tiny shuffle partitions.” Its normal signal is Task count decreases while throughput and result correctness remain acceptable. Here the requested behavior is “Shuffle partitions are badly sized at runtime due to variable daily input,” so that signal is not sufficient.

This action solves a different issue: “A batch runtime plan changes but analysts worry results are inconsistent.” Its normal signal is Both optimized and baseline execution return equivalent final records. Here the requested behavior is “Shuffle partitions are badly sized at runtime due to variable daily input,” so that signal is not sufficient.

Both optimized and baseline execution return equivalent final records pertains to a different control (Validate output correctness and compare adaptive physical plan independently) and a different need (A batch runtime plan changes but analysts worry results are inconsistent). It cannot establish that Enable supported Adaptive Query Execution partition coalescing met the stated goal.

That observation is a valid indicator for “One partition contains much more data than other shuffle partitions” when using Use supported AQE skew join handling and inspect partition statistics. The scenario instead tests “Shuffle partitions are badly sized at runtime due to variable daily input”; the relevant result is Runtime plan coalesces undersized shuffle partitions.

Task count decreases while throughput and result correctness remain acceptable pertains to a different control (Tune/adopt AQE coalescing and advisory partition settings) and a different need (Executors are overwhelmed by excessive task count from tiny shuffle partitions). It cannot establish that Enable supported Adaptive Query Execution partition coalescing met the stated goal.

This action solves a different issue: “Shuffle partitions are badly sized at runtime due to variable daily input.” Its normal signal is Runtime plan coalesces undersized shuffle partitions. Here the requested behavior is “One partition contains much more data than other shuffle partitions,” so that signal is not sufficient.

This action solves a different issue: “A batch runtime plan changes but analysts worry results are inconsistent.” Its normal signal is Both optimized and baseline execution return equivalent final records. Here the requested behavior is “One partition contains much more data than other shuffle partitions,” so that signal is not sufficient.

This action solves a different issue: “The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering.” Its normal signal is The final physical plan uses the better supported join for observed sizes. Here the requested behavior is “One partition contains much more data than other shuffle partitions,” so that signal is not sufficient.

This signal belongs to the adjacent case “A batch runtime plan changes but analysts worry results are inconsistent,” whose remedy is Validate output correctness and compare adaptive physical plan independently. The observed result for the present requirement is Straggler duration decreases without changing query semantics.

The final physical plan uses the better supported join for observed sizes pertains to a different control (Allow AQE to adapt supported join strategy at runtime) and a different need (The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering). It cannot establish that Use supported AQE skew join handling and inspect partition statistics met the stated goal.

Runtime plan coalesces undersized shuffle partitions would demonstrate the separate action Enable supported Adaptive Query Execution partition coalescing, which addresses “Shuffle partitions are badly sized at runtime due to variable daily input.” It does not verify the requirement “One partition contains much more data than other shuffle partitions” following Use supported AQE skew join handling and inspect partition statistics.

Use supported AQE skew join handling and inspect partition statistics belongs to the separate case “One partition contains much more data than other shuffle partitions,” where Straggler duration decreases without changing query semantics would be expected. That differs from the condition in this stem: “The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering.”

This action solves a different issue: “A batch runtime plan changes but analysts worry results are inconsistent.” Its normal signal is Both optimized and baseline execution return equivalent final records. Here the requested behavior is “The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering,” so that signal is not sufficient.

This action solves a different issue: “Shuffle partitions are badly sized at runtime due to variable daily input.” Its normal signal is Runtime plan coalesces undersized shuffle partitions. Here the requested behavior is “The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering,” so that signal is not sufficient.

This signal belongs to the adjacent case “One partition contains much more data than other shuffle partitions,” whose remedy is Use supported AQE skew join handling and inspect partition statistics. The observed result for the present requirement is The final physical plan uses the better supported join for observed sizes.

This signal belongs to the adjacent case “A batch runtime plan changes but analysts worry results are inconsistent,” whose remedy is Validate output correctness and compare adaptive physical plan independently. The observed result for the present requirement is The final physical plan uses the better supported join for observed sizes.

Task count decreases while throughput and result correctness remain acceptable would demonstrate the separate action Tune/adopt AQE coalescing and advisory partition settings, which addresses “Executors are overwhelmed by excessive task count from tiny shuffle partitions.” It does not verify the requirement “The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering” following Allow AQE to adapt supported join strategy at runtime.

The alternative Enable supported Adaptive Query Execution partition coalescing has a distinct purpose—Shuffle partitions are badly sized at runtime due to variable daily input. It would not establish the required outcome (Executors are overwhelmed by excessive task count from tiny shuffle partitions); the necessary verification concerns Task count decreases while throughput and result correctness remain acceptable.

The alternative Validate output correctness and compare adaptive physical plan independently has a distinct purpose—A batch runtime plan changes but analysts worry results are inconsistent. It would not establish the required outcome (Executors are overwhelmed by excessive task count from tiny shuffle partitions); the necessary verification concerns Task count decreases while throughput and result correctness remain acceptable.

Allow AQE to adapt supported join strategy at runtime is appropriate when “The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering”; it would be verified by The final physical plan uses the better supported join for observed sizes. It does not directly satisfy “Executors are overwhelmed by excessive task count from tiny shuffle partitions,” whose relevant acceptance evidence is Task count decreases while throughput and result correctness remain acceptable.

The final physical plan uses the better supported join for observed sizes pertains to a different control (Allow AQE to adapt supported join strategy at runtime) and a different need (The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering). It cannot establish that Tune/adopt AQE coalescing and advisory partition settings met the stated goal.

That observation is a valid indicator for “Shuffle partitions are badly sized at runtime due to variable daily input” when using Enable supported Adaptive Query Execution partition coalescing. The scenario instead tests “Executors are overwhelmed by excessive task count from tiny shuffle partitions”; the relevant result is Task count decreases while throughput and result correctness remain acceptable.

Both optimized and baseline execution return equivalent final records pertains to a different control (Validate output correctness and compare adaptive physical plan independently) and a different need (A batch runtime plan changes but analysts worry results are inconsistent). It cannot establish that Tune/adopt AQE coalescing and advisory partition settings met the stated goal.

Use supported AQE skew join handling and inspect partition statistics belongs to the separate case “One partition contains much more data than other shuffle partitions,” where Straggler duration decreases without changing query semantics would be expected. That differs from the condition in this stem: “A batch runtime plan changes but analysts worry results are inconsistent.”

This action solves a different issue: “Shuffle partitions are badly sized at runtime due to variable daily input.” Its normal signal is Runtime plan coalesces undersized shuffle partitions. Here the requested behavior is “A batch runtime plan changes but analysts worry results are inconsistent,” so that signal is not sufficient.

The alternative Allow AQE to adapt supported join strategy at runtime has a distinct purpose—The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering. It would not establish the required outcome (A batch runtime plan changes but analysts worry results are inconsistent); the necessary verification concerns Both optimized and baseline execution return equivalent final records.

That observation is a valid indicator for “The planner initially picks a shuffle join but one input becomes broadcast-eligible after filtering” when using Allow AQE to adapt supported join strategy at runtime. The scenario instead tests “A batch runtime plan changes but analysts worry results are inconsistent”; the relevant result is Both optimized and baseline execution return equivalent final records.

That observation is a valid indicator for “Executors are overwhelmed by excessive task count from tiny shuffle partitions” when using Tune/adopt AQE coalescing and advisory partition settings. The scenario instead tests “A batch runtime plan changes but analysts worry results are inconsistent”; the relevant result is Both optimized and baseline execution return equivalent final records.

That observation is a valid indicator for “Shuffle partitions are badly sized at runtime due to variable daily input” when using Enable supported Adaptive Query Execution partition coalescing. The scenario instead tests “A batch runtime plan changes but analysts worry results are inconsistent”; the relevant result is Both optimized and baseline execution return equivalent final records.