For a daily batch processing workload requiring hundreds of compute instances for a few hours, which is not time-sensitive and aims for cost reduction, what is the optimal strategy for resource utilization and managing preemptible VMs on Google Cloud Platform?
