QuestionQ36

Perform database monitoring, tuning, and optimization

A Snowflake customer is incurring higher-than-expected costs while migrating data-warehouse workloads from on premises to Snowflake. The migration workloads run on one warehouse and consist of many small INSERT statements rather than bulk-loading large extracts. That warehouse is configured as a single 2XL cluster because many parallel INSERT statements are scheduled during nightly loads.

How can the Administrator lower costs while minimizing the overall time required to load data-warehouse history?

Explanation

For high-concurrency, short-running work, a multi-cluster warehouse increases throughput by adding compute clusters, whereas enlarging a single cluster mainly increases resources available to individual queries. Using Small or X-Small clusters reduces the per-cluster credit rate, and multi-cluster scaling supplies additional clusters only when concurrent load demand requires them. Snowflake also notes that larger warehouses can consume more credits without improving performance when the workload is not a large concurrent bulk load.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!