Snowflake SnowPro Core Practice Test 88
Question 870
For an original SnowPro Core educational scenario, consider: A team configures a one-minute target lag for expensive multi-stage work without checking compute cost or source update frequency. Select TWO complementary steps: the targeted correction and its strongest prechange validation.
- Change: Evaluate supported Query Acceleration Service on eligible queries using its estimate and cost controls.
- Change: Review dynamic table refresh history, eligibility and resource use, then set an achievable target lag for business freshness needs.
- Check: Compare actual refresh lag, failures, dependency graphs and credits against required reporting SLA.
- Treat a declared TARGET_LAG as a hard contractual guarantee that no row can be late.
- Check: Compare average execution versus queue time and cluster saturation during a simulated peak.
Explanation
Correct Answer
B:Dynamic table target lag guides managed refresh behavior but does not promise zero-latency propagation or free compute. Evidence: Compare actual refresh lag, failures, dependency graphs and credits against required reporting SLA.
C:Dynamic table target lag guides managed refresh behavior but does not promise zero-latency propagation or free compute. Required intervention: Review dynamic table refresh history, eligibility and resource use, then set an achievable target lag for business freshness needs.
Incorrect Answers
A:Change: Evaluate supported Query Acceleration Service on eligible queries using its estimate and cost controls. A team configures a one-minute target lag for expensive multi-stage work without checking compute cost or source update frequency.
D:For this select two assessment, the shortcut changes or bypasses the necessary correctness, governance or operational boundary rather than correcting it. Dynamic table target lag guides managed refresh behavior but does not promise zero-latency propagation or free compute.
E:Check: Compare average execution versus queue time and cluster saturation during a simulated peak. Dynamic table target lag guides managed refresh behavior but does not promise zero-latency propagation or free compute.
Question 871
A Snowflake engineer is examining this platform situation: A data scientist calls collect on tens of millions of Snowpark rows before filtering in local Python. A proposed shortcut is: Download the entire warehouse dataset to a laptop as the default first step in every Snowpark query. What should the Snowflake engineer recommend as the narrowly scoped correction? Select ONE best answer.
- Match SELECT expressions to GROUP BY grain and use correct distinct or preaggregation strategy when source cardinality requires it.
- Download the entire warehouse dataset to a laptop as the default first step in every Snowpark query.
- Evaluate a supported materialized view for a repeatable eligible query with justified refresh and storage cost.
- Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps.
Explanation
Correct Answer
D:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. This targeted change addresses the stated technical or governance gap without discarding necessary controls.
Incorrect Answers
A:This option addresses a different Snowflake technical topic (Aggregations and grouping correctness) rather than the documented Snowpark pushdown and Python UDF boundaries decision.
B:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. For this best change assessment, the shortcut changes or bypasses the necessary correctness, governance or operational boundary rather than correcting it.
C:Evaluate a supported materialized view for a repeatable eligible query with justified refresh and storage cost. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Question 872
A Snowflake implementation review finds: A data scientist calls collect on tens of millions of Snowpark rows before filtering in local Python. The team proposes: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps. Which platform principle most strongly supports the recommended response? Select ONE best answer.
- Local or remote spill indicates execution pressure and can be affected by memory limits, join cardinality and data distribution.
- High concurrency can cause queuing independently of individual query efficiency; multi-cluster compute targets concurrent demand.
- Snowflake may reuse prior query results subject to eligibility and validity conditions; this cache is different from a warehouse's local data cache.
- Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Explanation
Correct Answer
D:Demonstrable success requires: The pipeline processes supported data within Snowflake while downloading only necessary results. This principle explains the proposed technical approach, which is: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps.
Incorrect Answers
A:This option addresses a different Snowflake technical topic (Query profile spilling and repartitioning) rather than the documented Snowpark pushdown and Python UDF boundaries decision.
B:High concurrency can cause queuing independently of individual query efficiency; multi-cluster compute targets concurrent demand. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
C:Snowflake may reuse prior query results subject to eligibility and validity conditions; this cache is different from a warehouse's local data cache. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Question 873
The data platform team has not yet approved changes. Current evidence indicates: A data scientist calls collect on tens of millions of Snowpark rows before filtering in local Python. Which evidence best tests the technical diagnosis before approving a disruptive change? Select ONE best answer.
- Inspect query profile scanned partitions, predicate form and representative date distribution.
- Compare query IDs, duration, queue time, bytes scanned, spill and credits over equivalent workloads.
- Compare query profile, clustering depth, maintenance credits and representative workload results.
- Compare generated SQL plans, data transfer volumes and managed package availability for a sample transform.
Explanation
Correct Answer
D:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. Confirm afterward that: The pipeline processes supported data within Snowflake while downloading only necessary results.
Incorrect Answers
A:Inspect query profile scanned partitions, predicate form and representative date distribution. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
B:Compare query IDs, duration, queue time, bytes scanned, spill and credits over equivalent workloads. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
C:Compare query profile, clustering depth, maintenance credits and representative workload results. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Question 874
A candidate correction has been tested and the platform team requires a post-change acceptance check. Original problem: A data scientist calls collect on tens of millions of Snowpark rows before filtering in local Python. The service owner needs positive and negative results and evidence that legitimate data workflows still function. Which acceptance result demonstrates correct platform behavior without disrupting authorized workflows? Select ONE best answer.
- Recurring reads improve without unacceptable data freshness or maintenance burden.
- Each group has the intended ranked or rolling values while original row-level detail is preserved.
- The pipeline processes supported data within Snowflake while downloading only necessary results.
- The query reads fewer relevant partitions while returning identical filtered rows.
Explanation
Correct Answer
C:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. The result must be validated using the documented dataset or access context.
Incorrect Answers
A:Recurring reads improve without unacceptable data freshness or maintenance burden. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
B:Each group has the intended ranked or rolling values while original row-level detail is preserved. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
D:The query reads fewer relevant partitions while returning identical filtered rows. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Question 875
An implementation sign-off request identifies the following problem: A data scientist calls collect on tens of millions of Snowpark rows before filtering in local Python. Which paired technical change and validation should close the implementation review? Select ONE best answer.
- Change: Evaluate supported Query Acceleration Service on eligible queries using its estimate and cost controls. Verify: Inspect query eligibility and acceleration history plus incremental service credits for representative workloads.
- Change: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps. Verify: Compare generated SQL plans, data transfer volumes and managed package availability for a sample transform.
- Change: Use appropriate window functions with PARTITION BY and ordered frames for row-level ranking or rolling calculations. Verify: Test row_number or moving-average output for customers with multiple timestamps and ties.
- Change: Use an appropriate MERGE with deterministic matched and unmatched clauses and validated source uniqueness. Verify: Test source-key duplicates, matched row counts and idempotent replay of the same batch.
Explanation
Correct Answer
B:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. The change is successful only when: The pipeline processes supported data within Snowflake while downloading only necessary results.
Incorrect Answers
A:Change: Evaluate supported Query Acceleration Service on eligible queries using its estimate and cost controls. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
C:Change: Use appropriate window functions with PARTITION BY and ordered frames for row-level ranking or rolling calculations. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
D:Change: Use an appropriate MERGE with deterministic matched and unmatched clauses and validated source uniqueness. A data scientist calls collect on tens of millions of Snowpark rows before filtering in local Python.
Question 876
A Snowflake engineer is examining this platform situation: A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls. A proposed shortcut is: Download the entire warehouse dataset to a laptop as the default first step in every Snowpark query. What should the Snowflake engineer recommend as the narrowly scoped correction? Select ONE best answer.
- Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps.
- Use appropriate window functions with PARTITION BY and ordered frames for row-level ranking or rolling calculations.
- Evaluate Snowflake Search Optimization Service for supported highly selective predicates and scoped table columns.
- Download the entire warehouse dataset to a laptop as the default first step in every Snowpark query.
Explanation
Correct Answer
A:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. Applied case: A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls.
Incorrect Answers
B:Use appropriate window functions with PARTITION BY and ordered frames for row-level ranking or rolling calculations. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
C:Evaluate Snowflake Search Optimization Service for supported highly selective predicates and scoped table columns. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
D:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. Use this instead: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps.
Question 877
A Snowflake implementation review finds: A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls. The team proposes: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps. Which platform principle most strongly supports the recommended response? Select ONE best answer.
- Snowflake may reuse prior query results subject to eligibility and validity conditions; this cache is different from a warehouse's local data cache.
- High concurrency can cause queuing independently of individual query efficiency; multi-cluster compute targets concurrent demand.
- TRY-style conversions can yield NULL for invalid input instead of stopping an entire transform, but silently ignoring invalid values is still a quality risk.
- Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Explanation
Correct Answer
D:Demonstrable success requires: The pipeline processes supported data within Snowflake while downloading only necessary results. Applied case: A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls.
Incorrect Answers
A:Snowflake may reuse prior query results subject to eligibility and validity conditions; this cache is different from a warehouse's local data cache. A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls.
B:High concurrency can cause queuing independently of individual query efficiency; multi-cluster compute targets concurrent demand. A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls.
C:TRY-style conversions can yield NULL for invalid input instead of stopping an entire transform, but silently ignoring invalid values is still a quality risk. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Question 878
The data platform team has not yet approved changes. Current evidence indicates: A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls. Which evidence best tests the technical diagnosis before approving a disruptive change? Select ONE best answer.
- Compare query IDs, duration, queue time, bytes scanned, spill and credits over equivalent workloads.
- Compare generated SQL plans, data transfer volumes and managed package availability for a sample transform.
- Inspect query eligibility and acceleration history plus incremental service credits for representative workloads.
- Count conversion failures, review source examples and reconcile accepted versus quarantined records.
Explanation
Correct Answer
B:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. This check tests the actual data, compute or access boundary before an operationally risky change.
Incorrect Answers
A:Compare query IDs, duration, queue time, bytes scanned, spill and credits over equivalent workloads. A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls.
C:Inspect query eligibility and acceleration history plus incremental service credits for representative workloads. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
D:Count conversion failures, review source examples and reconcile accepted versus quarantined records. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Question 879
A candidate correction has been tested and the platform team requires a post-change acceptance check. Original problem: A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls. The service owner needs positive and negative results and evidence that legitimate data workflows still function. Which acceptance result demonstrates correct platform behavior without disrupting authorized workflows? Select ONE best answer.
- Targeted point-lookups accelerate while maintenance costs remain proportional to business value.
- User-facing latency improves by reducing wait time while maintaining correctness and reasonable credits.
- The join yields expected grain and aggregate totals reconcile to source transactions.
- The pipeline processes supported data within Snowflake while downloading only necessary results.
Explanation
Correct Answer
D:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. The approved implementation step was: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps.
Incorrect Answers
A:Targeted point-lookups accelerate while maintenance costs remain proportional to business value. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
B:User-facing latency improves by reducing wait time while maintaining correctness and reasonable credits. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
C:The join yields expected grain and aggregate totals reconcile to source transactions. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
Question 880
For an original SnowPro Core educational scenario, consider: A developer assumes every arbitrary Python package runs inside standard SQL functions without dependency or runtime controls. Select TWO complementary steps: the targeted correction and its strongest prechange validation.
- Change: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps.
- Check: Inspect query profile scanned partitions, predicate form and representative date distribution.
- Change: Evaluate a supported materialized view for a repeatable eligible query with justified refresh and storage cost.
- Download the entire warehouse dataset to a laptop as the default first step in every Snowpark query.
- Check: Compare generated SQL plans, data transfer volumes and managed package availability for a sample transform.
Explanation
Correct Answer
A:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. Acceptance: The pipeline processes supported data within Snowflake while downloading only necessary results.
E:Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk. Required intervention: Express supported filters and transformations as Snowpark operations and use governed execution runtimes for Python-specific steps.
Incorrect Answers
B:Check: Inspect query profile scanned partitions, predicate form and representative date distribution. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
C:Change: Evaluate a supported materialized view for a repeatable eligible query with justified refresh and storage cost. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
D:For this select two assessment, the shortcut changes or bypasses the necessary correctness, governance or operational boundary rather than correcting it. Snowpark can push supported operations to Snowflake compute; premature data collection or unsupported runtime assumptions add cost and risk.
