ADA-C01 Question 413
Single answerCalculate and minimize IDLE time costA Snowflake administrator is reviewing warehouse spend for a reporting workload that runs in short bursts throughout the business day. Query History shows most statements complete in under 20 seconds, but Account Usage reveals that one warehouse accrues a large amount of IDLE time between bursts. The business wants to reduce compute cost without materially affecting report response times. Which configuration change is the BEST choice?
- A
Reduce the warehouse AUTO_SUSPEND setting to a low value such as 60 seconds and keep AUTO_RESUME enabled
- B
Increase the warehouse size so queued queries finish faster and the warehouse spends less time idle
- C
Disable AUTO_SUSPEND so the warehouse remains warm and avoids repeated start/stop activity
- D
Convert the warehouse to a multi-cluster warehouse in Maximized mode to eliminate idle credits
Show answer and explanation
Correct answer: A
Explanation
The key to minimizing IDLE time cost is to reduce the amount of time a warehouse remains running when no queries are executing. In Snowflake, warehouse billing continues while the warehouse is active, even if it is idle. For intermittent or bursty workloads, Snowflake best practice is to configure a short AUTO_SUSPEND interval and leave AUTO_RESUME enabled so the warehouse shuts down quickly after activity stops and starts again automatically for new queries. Administrators can analyze idle cost using views such as ACCOUNT_USAGE.WAREHOUSE_METERING_HISTORY, which separates compute usage into categories including idle time. Options that increase warehouse size or use multi-cluster behavior address performance and concurrency concerns, but they do not specifically solve idle billing for a warehouse that is simply left running between short query bursts.
- A. Correct.
Correct.
IDLEtime cost occurs when a virtual warehouse is running but not executing queries. For bursty workloads with gaps between short queries, lowering AUTO_SUSPEND reduces the amount of billed idle time by suspending the warehouse soon after activity stops. Keeping AUTO_RESUME enabled allows the warehouse to restart automatically when the next query arrives, which is the standard cost-optimization approach for intermittent workloads. - B. Incorrect.
Incorrect. Increasing warehouse size may reduce query runtime or queueing in some cases, but it does not directly address long periods where the warehouse is running with no work. In fact, a larger warehouse generally consumes credits at a higher rate, so idle periods can become even more expensive.
- C. Incorrect.
Incorrect. Disabling AUTO_SUSPEND is the opposite of what is typically recommended for reducing idle cost. While keeping the warehouse running can reduce the frequency of resumes, it causes the warehouse to continue billing during inactive periods, which is exactly the problem described in the scenario.
- D. Incorrect.
Incorrect. Multi-cluster warehouses are intended to address concurrency by adding clusters when needed. They do not eliminate idle billing, and in some cases can increase cost if not carefully configured. Maximized mode is designed to prioritize performance under concurrency, not to minimize idle time for a lightly bursty reporting workload.