Under the economy scaling policy, when will a multi-cluster Snowflake warehouse shut down?

• A. After 2 to 3 consecutive successful checks (performed at 1-minute intervals), which determine whether the load on the least-loaded cluster could be redistributed to the other clusters without spinning up the cluster again.
• B. After 5 to 6 consecutive successful checks (performed at 1-minute intervals), which determine whether the load on the least-loaded cluster could be redistributed to the other clusters without spinning up the cluster again.
• C. After 1 to 2 consecutive checks (performed at 1-minute intervals).
• D. After 7 to 8 consecutive checks (performed at 1-minute intervals).

Answer: (B)

Economy scaling is about saving costs by removing unused compute without hurting performance. In a multi-cluster warehouse, the system watches whether the workload on the least-loaded cluster can be absorbed by the other clusters without needing to spin up a new one. It checks this at one-minute intervals. If those checks stay true for five to six consecutive minutes, the system is confident enough to shut down that least-loaded cluster. This cadence provides a balance: it avoids turning off a cluster due to a brief lull, while still letting you reduce capacity when demand is consistently steady. Shorter streaks like one or two checks risk premature shutdown, while longer streaks like seven or eight checks would delay potential cost savings. Keep in mind the policy only applies when there are multiple clusters to reassign the load to.