COF-C03 - Snowflake AI Data Cloud Features and Architecture - Section 1.5

Configure virtual warehouses including size, multi-cluster warehouses, scaling policies, and auto-suspend and auto-resume.

Configure a virtual warehouse: choose a size for query complexity, enable multi-cluster warehouses for concurrency, and set the scaling policy (standard or economy). Use auto-suspend and auto-resume to control cost, and recognise how a warehouse caches data on its local disk while running.

virtual warehouse sizemulti-cluster warehousescaling policyauto-suspendauto-resume

Practice question for this objective

Free sampleSnowflake AI Data Cloud Features and Architecturemedium

A single analytical query joins several large tables and spills to remote storage because it lacks memory and local disk, so it runs slowly even when nothing else is on the warehouse. The team wants this one heavy query to finish faster by giving it more compute resources per cluster. Which change to the virtual warehouse best meets this need?

  • AConvert the warehouse to a multi-cluster warehouse with up to four clusters so additional clusters absorb the heavy query and shorten its runtime
  • BLower the auto-suspend timer to a few seconds so the warehouse restarts with a clean cache before each run of the slow analytical query
  • CSwitch the scaling policy from Standard to Economy so the warehouse conserves credits and devotes the saved compute to the slow analytical query
  • DIncrease the warehouse size to the next larger size so the query gets more memory and local disk per cluster and spills to remote storage less Correct
Recognise that scaling up to a larger warehouse size, not scaling out, gives a single heavy query more memory and local disk. A larger virtual warehouse size provisions more servers per cluster, which increases the memory and local SSD available to a single query so it spills less to remote storage and completes faster. Multi-cluster warehouses add clusters for concurrency, but one query still executes within a single cluster, so only a bigger size helps an isolated heavy query.

Why A is wrong: Multi-cluster scaling out adds clusters for concurrent queries, but a single query runs on one cluster only, so extra clusters do not give that query more memory or speed.

Why B is wrong: A shorter auto-suspend only changes when the warehouse idles off and would drop the cache, so it does not add resources or make the heavy query run any faster.

Why C is wrong: Scaling policy only governs when extra clusters start in a multi-cluster warehouse, so it adds no resources to a single query and the Economy claim is incorrect.

Why D is correct: Scaling up to a larger warehouse size adds compute, memory, and local disk to the single cluster the query uses, reducing spilling and shortening the heavy query's runtime.

See more COF-C03 practice questions, answers explained.

More in this domain

Back to all Snowflake AI Data Cloud Features and Architecture objectives, or the COF-C03 cert hub.

Examworthy is not affiliated with or endorsed by Snowflake. Original, blueprint-aligned practice material only.