To enhance the resilience of Compute Engine instances against crashes or system terminations, which configurations should be applied?