When using a PAC within a container, you should use autoscaling to ensure that scaling down your application's resources does not result in the unexpected termination of Enterprise Server work, particularly in the case of long-running batch jobs. By using autoscaling you can prevent new work from being allocated to a PAC, and also guarantee that the PAC and its member regions will remain running until it has completed its current workload. The auto-scaling rules that you specify should contain a grace period during which this shutdown procedure will be allowed to run before the host container is terminated. In the case where a PAC is running batch jobs, this grace period should be at least the expected duration of time for the batch job to complete.