ADA-C01 Question 411
Select 2Demonstrate cost saving strategiesA Snowflake administrator is asked to reduce monthly compute spend for a BI workload without noticeably affecting dashboard response times. The workload has these characteristics: dashboards run every 15 minutes from 7:00 AM to 7:00 PM, query patterns are highly repetitive, and warehouse monitoring shows frequent short idle periods between bursts of activity. The current solution uses a Medium warehouse with AUTO_SUSPEND = 600 seconds and no query acceleration features enabled. Which TWO actions would most directly reduce compute cost while preserving user experience?
- A
Reduce AUTO_SUSPEND to a much lower value, such as 60 seconds, so the warehouse stops billing sooner during repeated idle gaps
- B
Enable Query Acceleration Service on the warehouse to reduce total credits consumed for all dashboard queries
- C
Resize the warehouse from Medium to Small and evaluate whether query performance remains acceptable for the BI concurrency level
- D
Disable result caching so dashboards always execute against the warehouse and produce more predictable performance
- E
Convert the workload to use a Snowpark-optimized warehouse because BI dashboards benefit from specialized compute
Show answer and explanation
Correct answers: A, C
Explanation
The best answers are to lower AUTO_SUSPEND and rightsize the warehouse. Snowflake cost-saving best practices emphasize minimizing idle warehouse time and selecting the smallest warehouse size that still satisfies performance and concurrency requirements. In this scenario, the warehouse has frequent short idle periods, so a 600-second AUTO_SUSPEND setting likely wastes credits. Lowering it can reduce billed runtime without materially harming user experience. Also, repetitive BI workloads often run well on smaller warehouses, especially when supported by Snowflake caching behavior. By contrast, Query Acceleration Service is designed for performance improvement on eligible queries and adds cost; it is not a default cost-reduction strategy. Disabling result cache is counterproductive for repetitive dashboard queries because cached results can avoid compute entirely. Snowpark-optimized warehouses are specialized for different workload types and are not appropriate here. These recommendations align with Snowflake guidance around warehouse tuning, auto-suspend/auto-resume, and warehouse right-sizing as core administrative cost-control strategies.
- A. Correct.
Correct. Warehouses are billed while running, including idle time until suspension. For bursty BI workloads with short idle gaps, lowering AUTO_SUSPEND from 600 seconds to something like 60 seconds often reduces wasted compute credits significantly. Because the dashboards run at regular intervals and query patterns are repetitive, resume overhead is typically small compared with the cost savings from avoiding long idle periods.
- B. Incorrect.
Incorrect. Query Acceleration Service can improve performance for eligible queries by offloading portions of query processing, but it is an additional service that can increase cost rather than serving as a direct cost-reduction measure for all dashboard queries. It is primarily a performance optimization, not a general compute savings strategy.
- C. Correct.
Correct. Rightsizing the warehouse is a primary Snowflake cost optimization practice. If monitoring shows that a Medium warehouse is larger than needed for the dashboard workload, moving to a Small warehouse can cut compute cost while still meeting response time requirements. The key is to test concurrency and latency to verify that performance remains acceptable.
- D. Incorrect.
Incorrect. Disabling result caching would usually increase compute consumption because repetitive dashboard queries that could be served from the result cache would instead re-run on the warehouse. For highly repetitive BI workloads, result caching is one of the mechanisms that helps reduce compute usage while maintaining fast response times.
- E. Incorrect.
Incorrect. Snowpark-optimized warehouses are intended for workloads that need more memory and compute resources for certain Snowpark and ML-style operations. They are not a cost-saving choice for standard BI dashboard queries and would generally not be the appropriate optimization for this scenario.