Open in Playground Download design mesheryctl design import -f https://raw.githubusercontent.com/meshery/meshery.io/master/catalog/a7701717-2abc-4223-aab0-3eb09693561e/0.0.2/design.yml

What This Design Does:

A horizontally auto-scaling Kubernetes web application. A Deployment runs the web tier behind a ClusterIP Service, while a HorizontalPodAutoscaler dynamically adjusts replicas between 2 and 10 based on CPU utilization. All resources live in a dedicated Namespace. This pattern demonstrates Kubernetes-native elasticity with minimal component count: traffic scales with demand, failed pods are replaced automatically, and idle load collapses back to the two-replica floor. It runs on any conformant Kubernetes cluster, including EKS, GKE, AKS, kind, and minikube.

Design Snapshot

autoscaling-demo design snapshot

Caveats and Consideration:

Requires metrics-server. The HPA cannot function without metrics-server installed in the cluster, because it depends on the resource metrics API to read CPU utilization. Most managed clusters (EKS, GKE, AKS) include it by default; kind and minikube may require enabling the metrics-server addon manually. Without it, the HPA will report metrics and will not scale. The image is a placeholder. nginx:1.27-alpine is used to demonstrate the pattern. Replace it with your actual application image and adjust resources.requests.cpu accordingly — HPA CPU targets are calculated as a percentage of the requested CPU, so a missing or too-small request value will distort scaling behaviour. CPU is a proxy metric. This HPA scales on CPU utilization. Applications that are I/O-bound, queue-depth-bound, or memory-bound should use custom or external metrics instead of — or in addition to — CPU. The design is intentionally minimal; adding custom metrics requires Prometheus or another metrics adapter.

Compatibility:

Design Details

Related Designs

deployment
vault-0.30.1.tgz
MESHERY40c7
`