@@ -1671,7 +1671,7 @@ Increasing the warehouse size will not always mean an increase in performance an
- If refactoring has not improved the Model Efficiency, an increase to the Warehouse Size may be warranted.
- If Model Efficiency is Good but the model is reaching the timeout limit for the warehouse then an increase to the Warehouse Size may be warranted.
The [Snowflake warehouse sizing](/handbook/enterprise-data/platform/snowflake-warehouse-optimization/) handbook page has guidelines on properly sizing dbt models.
The [warehouse benchmarking](/handbook/enterprise-data/platform/snowflake-benchmarking-guide/) handbook page has guidelines on properly sizing dbt models.
Guidelines to choose the correct warehouse for a particular dbt model:
Precisely choosing the correct warehouse requires benchmarking different warehouses:
- The user should base the initial *estimated* warehouse size on the number of partitions scanned within the query
- Benchmark by increasing the warehouse size and comparing the 'cost' vs 'reduction in query time'
- In the context of the total credits consumed by the dbt DAG, right-sizing warehouses by model for the smallest models does not lead to much cost reduction
- In the context of the total credits consumed by the dbt DAG, right-sizing warehouses by model for the smallest models does not lead to much cost reduction.
