Snowflake Auto-Suspend Savings Estimator

Project how much Snowflake spend you can trim by reducing auto-suspend buffers. Enter your current and target delay settings along with warehouse cost to see monthly and annual savings, idle minutes eliminated, and the percentage reduction in waste.

Minutes Snowflake keeps the warehouse running after the last query today.
Minutes you plan to keep the warehouse warm before suspending.
Blended hourly cost for the virtual warehouse (including credits or commitments).
Optional — defaults to 18 idle periods per day when blank.
Optional — defaults to 22 billing days per month when blank.

Confirm Snowflake credit pricing and workload patterns with your FinOps team before adjusting production warehouses.

Examples

  • Example 1 — Current delay 10 minutes, target 2 minutes, $6.50/hour, 90 idle windows per day, 22 active days ⇒ Monthly savings: $1,716.00 | Annual savings: $20,592.00 | Idle minutes eliminated each month: 15,840 | Idle runtime cut: 80.00% | Remaining idle hours per month: 66.00
  • Example 2 — Current delay 6 minutes, target 3 minutes, $3.80/hour, 40 idle windows per day, 28 active days ⇒ Monthly savings: $212.80 | Annual savings: $2,553.60 | Idle minutes eliminated each month: 3,360 | Idle runtime cut: 50.00% | Remaining idle hours per month: 56.00

FAQ

How do multi-cluster warehouses factor in?

Run the calculation per cluster or multiply idle windows by the average cluster count active during idle periods to capture parallel warehouses.

Can I include resume charges?

Resume activity consumes credits just like normal compute time. Add any extra resume overhead to the hourly cost input to keep the savings conservative.

What if I automate suspends via tasks?

Task-driven suspends effectively shorten the delay. Enter the equivalent minutes saved to see how much a cron-based approach could offset spend.

Additional Information

  • Idle windows represent distinct lulls between query bursts. Multiply peak hours by typical pauses to estimate an accurate count.
  • Savings assume all idle windows complete without new queries. If workloads resume before suspension, actual savings will be lower.
  • Blend credit pricing, discounts, and committed spend into the hourly cost for a true cash projection.
  • Keeping a small buffer avoids cold-start latency; most teams land between 1 and 5 minutes for interactive workloads.