Explanation
Why the waste happens and who it affects.
This often results in redundant and underutilized warehouses, as workloads frequently do not require the full capacity of even the smallest warehouse size.
By consolidating compatible workloads onto shared warehouses, organizations can maximize utilization, reduce idle runtime across the fleet, and significantly lower total credit consumption. Cost allocation can still be approximated with the QUERY_ATTRIBUTION_HISTORY view, which attributes compute credits to individual queries but excludes warehouse idle time and very short queries (about 100 ms or less), so idle time must be allocated separately, for example proportionally.
Billing model
The pricing dimensions that drive this cost.
- Virtual warehouses consume Snowflake credits while running, at an hourly rate set by warehouse size; each size step up approximately doubles the credits billed per hour
- Credits are billed per second, with a 60-second minimum each time a warehouse starts or resumes; suspended warehouses consume no credits
- Each separate warehouse accrues its own running time, including idle time before auto-suspend and the minimum charge on every resume, so many lightly used warehouses consume more credits than fewer, well-utilized ones
How to detect
4 checks to find it in your estate.
- Review warehouse usage patterns to identify warehouses operating consistently below concurrency capacity (e.g., low simultaneous query counts).
- Evaluate whether workloads running on separate warehouses have compatible scheduling, priority, or SLA needs that allow consolidation.
- Assess existing query queuing or resource contention risks before consolidation to avoid performance degradation.
- Validate with workload owners that consolidation would not introduce unacceptable performance risks.
How to fix
4 ways to remove the waste.
- Consolidate compatible workloads onto shared warehouses to improve overall utilization without sacrificing performance.
- Adjust warehouse sizing or enable multi-cluster scaling if necessary to accommodate increased concurrency after consolidation.
- Validate SLA and performance expectations with all impacted business units or workload owners prior to consolidation.
- Use the QUERY_ATTRIBUTION_HISTORY view to allocate per-query costs across teams sharing a consolidated warehouse, and allocate warehouse idle time separately (for example, in proportion to attributed usage).
Documentation
Vendor references for pricing and configuration.