Your Compute Engine instance group is set to autoscale based on overall CPU usage, but your application experiences significant latency spikes before new instances are fully added. To maintain more consistent latency for end users, which two autoscaler configuration changes should you implement? (Select two.)