To avoid this issue, don't perform scaling operations that release more than XX instances in an hour. Every time you release an instance during a scale-down operation, the instance is rebooted to ensure the next App Service Plan can get a clean instance. When you perform too many scaling operations in quick succession, instance reboots can cause performance issues for other App Services. Therefore we intentionally put a throttling mechanism for scaling that prevents you from executing scaling operations more than the acceptable limit in quick succession.