gke-workload-scaling

Configure Horizontal Pod Autoscaling (HPA) for GKE deployments by applying a manifest that auto-scales pod replicas based on observed CPU, memory, or custom metrics using the provided example manifest.

Prerequisites for enabling Vertical Pod Autoscaling (VPA) on GKE workloads include having resource requests defined and the metrics server available. Proper VPA and autoscaler configuration is required to ensure safe scaling activities.

Use manual scaling for your Kubernetes deployment when you need immediate intervention or fixed replica counts for testing. HPA is suited for automatic scaling based on observed metrics, while manual scaling provides direct, immediate control.

GKE Cluster Autoscaler works with HPA and VPA by ensuring the node pool scales to support scaled workloads. You must consider node pool scaling to support increased pod counts and optimize resource usage during capacity planning.

Vertical Pod Autoscaling (VPA) automatically adjusts CPU and memory reservations for pods using recommended update modes. Templates are provided to configure VPA behavior for optimizing resource allocation in staging and production GKE environments.

You can use custom metrics for HPA autoscaling in GKE production environments. The manifest-based approach supports auto-scaling pod counts based on observed CPU, memory, or custom metrics to manage workload capacity effectively.