An EKS cluster using managed node groups hosts microservices. The team has installed the Kubernetes Metrics Server and wants Pods to autoscale based on a target CPU utilization percentage with minimal operational effort. What combination accomplishes this?

The correct solution involves two key components for efficient autoscaling with minimal operational effort. The Horizontal Pod Autoscaler (HPA) is essential for scaling Pods based on metrics like CPU utilization, directly addressing the requirement for Pods to autoscale based on target CPU. Since the Kubernetes Metrics Server is already installed, HPA can readily consume this data. The Cluster Autoscaler then ensures that new nodes are provisioned or deprovisioned as needed to accommodate the HPA's scaling decisions, maintaining optimal resource utilization. Auto-discovery simplifies its configuration for EKS managed node groups. Incorrect options: Modifying the Auto Scaling group directly for worker nodes based on group CPU utilization is less granular and doesn't directly scale individual Pods based on their specific CPU usage. While HPA is correct, VPA in recommender mode only provides recommendations and doesn't actively scale Pods or nodes, and VPA is typically for vertical scaling (resource requests/limits), not horizontal scaling based on CPU utilization. Running an SSM Automation document for manual node group size adjustments is not an automated or dynamic autoscaling solution.