ADA-C01 Question 405
Single answerManage costs for virtual warehousesA Snowflake administrator is reviewing compute spend and finds that a BI virtual warehouse named BI_WH incurs high costs overnight. Query history shows many short dashboard queries between 6:00 AM and 11:00 PM, but only a few sporadic queries after hours. Users also report that dashboards must remain responsive during business hours. The warehouse is currently configured as MEDIUM, AUTO_SUSPEND = 600, AUTO_RESUME = TRUE, and MIN_CLUSTER_COUNT = MAX_CLUSTER_COUNT = 1. Which change would best reduce compute cost while preserving the required user experience?
- A
Reduce AUTO_SUSPEND to 60 seconds so the warehouse stops sooner during low-activity periods while still automatically resuming when queries arrive
- B
Disable AUTO_RESUME so that users manually restart the warehouse only when needed, preventing unnecessary resumes overnight
- C
Increase the warehouse size to LARGE so queries finish faster and total credits consumed decrease
- D
Convert the warehouse to a multi-cluster warehouse with MAX_CLUSTER_COUNT = 3 to improve concurrency and reduce nightly costs
Show answer and explanation
Correct answer: A
Explanation
The most effective cost-control action is to reduce idle compute time while keeping the warehouse easy for BI tools and dashboard users to access. Snowflake best practices for warehouse cost management emphasize configuring AUTO_SUSPEND and AUTO_RESUME appropriately so warehouses do not run unnecessarily between bursts of activity. In this case, a 600-second auto-suspend setting leaves the warehouse running for long periods after brief overnight queries. Lowering AUTO_SUSPEND to 60 seconds is a practical optimization because it reduces wasted runtime while preserving seamless access through AUTO_RESUME. Alternatives such as disabling AUTO_RESUME harm usability, increasing warehouse size targets the wrong problem, and enabling multi-cluster addresses concurrency rather than idle-cost reduction. Relevant Snowflake documentation includes guidance on virtual warehouse cost control, warehouse billing behavior, and using AUTO_SUSPEND/AUTO_RESUME to minimize unnecessary compute consumption.
- A. Correct.
Correct. Snowflake bills warehouses per-second with a minimum charge when a warehouse starts or resumes. In this scenario, the key issue is that the warehouse stays running for up to 10 minutes after the last query because AUTO_SUSPEND is set to 600 seconds. Since overnight usage is only sporadic, lowering AUTO_SUSPEND to 60 seconds will reduce idle runtime and therefore lower compute cost. AUTO_RESUME = TRUE preserves the user experience because queries can still start the warehouse automatically when needed.
- B. Incorrect.
Incorrect. Disabling AUTO_RESUME may reduce some automatic restarts, but it creates a poor user experience because dashboard users would encounter failures or delays until someone manually resumes the warehouse. The scenario explicitly requires dashboards to remain responsive during business hours. This option shifts operational burden to users and administrators rather than optimizing cost efficiently.
- C. Incorrect.
Incorrect. Increasing the warehouse to LARGE generally increases the credit consumption rate per unit of runtime. While larger warehouses can complete some workloads faster, that does not make this the best choice here because the main identified waste is idle time overnight, not insufficient compute power during query execution. Upsizing would likely increase cost rather than reduce it.
- D. Incorrect.
Incorrect. Multi-cluster warehouses are designed to handle concurrency by adding clusters when many queries run at the same time. The scenario does not indicate concurrency bottlenecks overnight; instead, it describes sparse after-hours usage. Adding multi-cluster capability would not address the idle-cost problem and could increase spending if additional clusters are used.