ARA-C01 Question 431
Single answerResource monitoringA Snowflake architect supports a shared analytics environment with separate virtual warehouses for ETL, BI, and ad hoc data science workloads. Near the end of each month, the data science warehouse frequently consumes far more credits than expected, causing budget overruns before the finance team notices. The architect must reduce the risk of runaway spend while minimizing disruption to scheduled ETL and dashboard workloads. Which approach BEST addresses this requirement?
- A
Create a resource monitor for the data science warehouse with multiple credit thresholds and actions, such as notifying at lower percentages and suspending the warehouse when the monthly quota is reached.
- B
Enable auto-suspend on the data science warehouse and rely on warehouse resizing policies to prevent the warehouse from exceeding the monthly budget.
- C
Create one account-level resource monitor that immediately suspends all warehouses once total account usage reaches the monthly budget.
- D
Use query acceleration service on the data science warehouse so long-running queries finish faster and therefore cannot exceed the monthly credit allocation.
Show answer and explanation
Correct answer: A
Explanation
The best answer is to use a warehouse-specific resource monitor with staged actions. Snowflake resource monitors are intended to control credit consumption by setting quotas for a specified interval and defining actions at thresholds, such as sending notifications and suspending assigned warehouses. Because the issue is isolated to the ad hoc data science workload, targeting only that warehouse aligns with the requirement to minimize disruption to ETL and BI operations. Best practice is to use progressive thresholds, for example notify at lower percentages and suspend at 100% of the monthly quota, so stakeholders can react before spend becomes excessive. In contrast, auto-suspend reduces idle compute but does not cap monthly spend; account-level suspension is too disruptive for this use case; and performance features such as Query Acceleration Service are not substitutes for credit governance. See Snowflake documentation on resource monitors, warehouse assignment, monitor schedules, threshold actions, and suspend behavior for implementation details.
- A. Correct.
Correct. Resource monitors are designed to track credit usage and trigger actions at defined thresholds. In this scenario, assigning a monitor specifically to the data science warehouse allows the architect to target the unpredictable workload without affecting ETL and BI warehouses. Using multiple thresholds for notifications and a suspend action at the quota helps both finance visibility and cost control. This is the most appropriate way to limit runaway spend for a specific warehouse while minimizing impact on other workloads.
- B. Incorrect.
Incorrect. Auto-suspend helps reduce waste from idle warehouses, and resizing can affect performance and cost rate, but neither enforces a monthly credit budget. A warehouse can still consume significant credits while actively running queries. This option reflects a common misconception that operational warehouse settings replace budget enforcement, but resource monitors are the Snowflake feature intended for credit consumption governance.
- C. Incorrect.
Incorrect. An account-level resource monitor can be used to control overall spending, but immediately suspending all warehouses would disrupt ETL and BI workloads, which the scenario explicitly says should be minimized. This approach is overly broad and does not isolate the problematic workload. It is plausible because it addresses budget overruns, but it is not the best fit for the stated business requirement.
- D. Incorrect.
Incorrect. Query Acceleration Service can improve performance for certain eligible queries, but it is not a budget control mechanism and may itself add compute cost. Faster query completion does not guarantee lower overall credit consumption in a way that enforces a monthly spending cap. This option confuses performance optimization with spend governance.