COF-C03 Question 65
Single answerScaling policiesA data engineering team uses a multi-cluster virtual warehouse to support a dashboard application with highly variable concurrency throughout the day. During peak hours, many short-running queries queue up, but overnight the workload is light. The team wants to reduce query queuing during spikes while avoiding unnecessary compute cost when demand drops. Which warehouse configuration best meets these requirements?
- A
Configure the warehouse as multi-cluster with auto-scale mode and set the scaling policy to STANDARD
- B
Configure the warehouse as multi-cluster with maximized mode and set the scaling policy to ECONOMY
- C
Configure the warehouse as a single-cluster warehouse and set the scaling policy to STANDARD
- D
Configure the warehouse as multi-cluster with auto-scale mode and set the scaling policy to ECONOMY
Show answer and explanation
Correct answer: A
Explanation
Snowflake multi-cluster warehouses are designed to address concurrency by adding clusters when query load increases. Auto-scale mode starts and stops clusters automatically based on workload, which is appropriate for environments with variable demand. The scaling policy controls how aggressively Snowflake starts additional clusters: STANDARD favors performance and faster responsiveness to concurrency increases, while ECONOMY is more conservative and can tolerate more queueing to reduce cost. Maximized mode is generally used when all configured clusters should run continuously, which does not match a workload that is only busy during peak hours. According to Snowflake documentation and best practices, choose auto-scale multi-cluster warehouses for fluctuating concurrency and use STANDARD when minimizing queueing is more important than minimizing cluster start frequency.
- A. Correct.
Correct. A multi-cluster warehouse in auto-scale mode can automatically start additional clusters when concurrency increases and shut them down when demand falls, which directly addresses intermittent query queuing and cost control. The STANDARD scaling policy is more responsive to starting additional clusters, helping reduce queuing during spikes. This is the best fit when performance during peak concurrency is the priority, while still allowing scale-down during quieter periods.
- B. Incorrect.
Incorrect. Maximized mode keeps all clusters running regardless of actual demand, which increases cost and does not align with the requirement to avoid unnecessary compute usage overnight. In addition, the scaling policy setting is not the main cost-control mechanism here because maximized mode intentionally runs at the maximum cluster count.
- C. Incorrect.
Incorrect. A single-cluster warehouse cannot add clusters to handle concurrent query spikes, so it will not effectively reduce queueing caused by high concurrency. The scaling policy applies to multi-cluster warehouse behavior, so this option reflects a common misconception that scaling policy alone can help a single-cluster warehouse handle concurrency bursts.
- D. Incorrect.
Incorrect. Auto-scale mode is appropriate, but the ECONOMY scaling policy is less aggressive about adding clusters. That can reduce costs, but it may allow more queuing during peak periods compared with STANDARD. Since the requirement emphasizes reducing query queuing during spikes while still scaling down later, STANDARD is the better choice.