Your company uses Google Kubernetes Engine (GKE) as a platform for all workloads. Your company has a single large GKE cluster that contains batch, stateful, and stateless workloads. The GKE cluster is configured with a single node pool with 200 nodes. Your company needs to reduce the cost of this cluster but does not want to compromise availability. What should you do?

The correct option leverages Horizontal Pod Autoscaling (HPA) and node autoscaling to optimize resource utilization and reduce costs without compromising availability. HPA scales pods based on demand, ensuring resources are only consumed when needed. Node autoscaling then adjusts the number of nodes to match the pod requirements, preventing over-provisioning. This dynamic scaling minimizes idle resources, directly reducing costs. Creating a second cluster for batch workloads (option 1) might improve isolation but doesn't inherently reduce costs; it could even increase management overhead. Allocating existing nodes across two clusters might lead to underutilization in one or both. Configuring CPU and memory limits (option 2) is good practice for resource management but doesn't automatically scale resources down when not needed, so it won't reduce costs as effectively as autoscaling. Using preemptible VMs (option 4) significantly reduces costs but compromises availability, as preemptible VMs can be terminated at any time, making them unsuitable for workloads requiring high availability.