ADA-C01 Question 222
Single answerImplement and maintain a scaling policyA Snowflake administrator manages a multi-cluster virtual warehouse named ETL_WH that supports a critical ELT workload. During hourly load windows, dozens of concurrent transformation queries arrive at the same time, causing queued statements and missed SLAs. Outside those windows, the warehouse is lightly used and cost control is important. The administrator wants Snowflake to automatically add compute only when concurrency increases and reduce compute when demand drops, while minimizing unnecessary cluster usage. Which warehouse configuration best meets these requirements?
- A
Configure ETL_WH as a multi-cluster warehouse with MIN_CLUSTER_COUNT = 1, MAX_CLUSTER_COUNT greater than 1, and SCALING_POLICY = STANDARD
- B
Configure ETL_WH as a multi-cluster warehouse with MIN_CLUSTER_COUNT = MAX_CLUSTER_COUNT and SCALING_POLICY = ECONOMY
- C
Keep ETL_WH as a single-cluster warehouse and set AUTO_SUSPEND to a low value so queued queries clear more quickly
- D
Configure ETL_WH as a multi-cluster warehouse with MIN_CLUSTER_COUNT = 0, MAX_CLUSTER_COUNT greater than 1, and SCALING_POLICY = STANDARD
Show answer and explanation
Correct answer: A
Explanation
For concurrency-driven workloads, the correct design is to use a multi-cluster warehouse so Snowflake can add clusters when many queries are submitted at once. The scaling policy determines how aggressively Snowflake starts and stops clusters. STANDARD is better for performance-sensitive workloads because it is more proactive in adding clusters to reduce queueing, whereas ECONOMY is more conservative and prioritizes cost efficiency over immediate responsiveness. To allow dynamic scaling while avoiding excess baseline cost, administrators typically set MIN_CLUSTER_COUNT to 1 and MAX_CLUSTER_COUNT to the upper limit needed for peak concurrency. This aligns with Snowflake best practices for implementing and maintaining a scaling policy: use multi-cluster warehouses for concurrency, tune min/max clusters based on observed demand, and choose STANDARD versus ECONOMY according to the balance between SLA requirements and cost optimization. Relevant Snowflake documentation covers multi-cluster warehouses, warehouse scaling policies, and warehouse parameter behavior such as AUTO_SUSPEND.
- A. Correct.
Correct. A multi-cluster warehouse can automatically start additional clusters to handle bursts of concurrent queries and reduce queuing. Setting MIN_CLUSTER_COUNT = 1 keeps one cluster available during active periods, while MAX_CLUSTER_COUNT greater than 1 allows Snowflake to scale out when concurrency rises. The STANDARD scaling policy is designed to be more responsive in starting additional clusters, which is appropriate for SLA-sensitive workloads where minimizing queue time is more important than minimizing cluster starts.
- B. Incorrect.
Incorrect. If MIN_CLUSTER_COUNT = MAX_CLUSTER_COUNT, the warehouse effectively runs at a fixed number of clusters and does not scale dynamically within that range. That reduces the benefit of automatic scaling based on workload. In addition, ECONOMY delays adding clusters more than STANDARD, which can increase queueing for a time-sensitive ELT workload. This option may control cost, but it does not best meet the stated requirement to add compute only as needed while still protecting SLAs.
- C. Incorrect.
Incorrect. AUTO_SUSPEND controls when an idle warehouse is suspended; it does not address query concurrency on an active warehouse. A single-cluster warehouse cannot scale out horizontally to reduce queueing caused by many simultaneous queries. This is a common misconception: suspension settings affect idle cost, not active concurrency capacity.
- D. Incorrect.
Incorrect. Snowflake does not allow MIN_CLUSTER_COUNT to be set to 0 for a multi-cluster warehouse. The minimum cluster count must be at least 1. Although STANDARD is an appropriate scaling policy for responsiveness, this configuration itself is invalid.