Snowflake Warehouse

Snowflake Warehouse recommends how to size your Snowflake warehouses, based on their average running time and average queue size, so you can right-size compute instead of guessing.

Navigating to Snowflake Warehouse

Navigate to Compute.
From any Snowflake data source, select Warehouse.

The Warehouse page has two tabs: Warehouse Recommendations and Warehouse Utilization.

Filters

The Data Source Filter switches between connected Snowflake accounts. Organization Unit and Cost Center filters are also available to refine the graphs, dashboards, and tables on this page.

Warehouse Recommendation is a quadrant chart plotting Average Running Time Ratio (x-axis) against Average Queue Load Ratio (y-axis) across all your warehouses. If you have the appropriate role permissions, select Edit Configurations to adjust the Expected Warehouse Latency and Expected Warehouse Queue Load values that define the chart's default thresholds.

The four quadrants:

Quadrant	Running Time	Queue Load
Underutilized	Low	Low
Ineffective	High	Low
Effective	Low	High
Overloaded	High	High

Based on which quadrant a warehouse falls into, ADOC recommends one of:

Upscale — for warehouses with high load and high running time.
Downscale — for idle warehouses, or those with low running time and low load.
Merge — for a warehouse with high load, routing some of its queries to another warehouse instead.

Applying a recommendation shows the cost of your current warehouse size alongside the projected cost at the recommended size.

Warehouse Sizing Recommendation is a table with the same underlying recommendation logic, one row per warehouse.

Column	Description
Warehouse Name	The warehouse's name.
Warehouse Size	Its current size.
Average Running Time / Average Queue Load	The values driving the recommendation.
Recommended	Recommendations made out of total evaluations for this warehouse.
Active Since	How long the warehouse has been operational.
Recommendation	Merge, Upscale, or Downscale.
Action	Available only for Upscale and Downscale recommendations — applies the resize directly.

You can select a new warehouse size directly from this table.

Warehouse Utilization

Warehouse Wastage plots each warehouse's wastage percentage (x-axis) against its productivity percentage (y-axis) for the selected time range.

Warehouse Busy State — actively processing tasks.
Warehouse Idle State — operational but not processing tasks.
Warehouse Suspend State — temporarily halted, typically to save cost.

Calculations:

Wastage % = (Idle Time × 100) / (Idle Time + Busy Time)
Productivity % = (Busy Time × 100) / (Idle Time + Busy Time)

Warehouses in the chart's red section have extended idle periods and are worth investigating. Select a warehouse from the dropdown for a detailed view.

Recommendation: Keep wastage below 5%. If it's higher, reduce the Auto Suspend timeout. If it's consistently low, increasing the Auto Suspend timeout can reduce transition costs (the overhead of repeatedly suspending and resuming).

Warehouse Performance (Top 25 Idle Warehouses) shows the 25 warehouses with the most idle time, broken down into idle, busy, and suspend time (yellow, blue, and gray respectively). Hover over a bar for the exact duration in each state, and the resulting productivity and wastage percentages.

What's next

Snowflake Costs – See the cost impact of warehouse sizing.
Snowflake Warehouse – Cross-reference warehouse load against query execution data.

Last updated on

Was this page helpful?