How to provision a Google GKE cluster and deploy a workload

Provisioning a Google GKE cluster

Google Kubernetes Engine (GKE) is Google Cloud's managed Kubernetes, the original home of the project. GKE offers two modes: Autopilot, which manages nodes for you and bills per pod, and Standard, which gives you full control of node pools.

Prerequisites

• A Google Cloud project with billing enabled.
• The gcloud CLI and kubectl installed.

Steps

1. Set the project and enable APIs

gcloud config set project my-project
gcloud services enable container.googleapis.com

2. Choose Autopilot or Standard

Pick Autopilot for a hands-off, pod-billed experience; pick Standard when you need specific machine types, GPUs, or fine node control.

3. Create the cluster

Standard mode with autoscaling:

gcloud container clusters create demo \
  --region us-central1 \
  --num-nodes 1 \
  --enable-autoscaling --min-nodes 1 --max-nodes 5

For Autopilot: gcloud container clusters create-auto demo --region us-central1.

4. Get cluster credentials

gcloud container clusters get-credentials demo --region us-central1
kubectl get nodes

5. Deploy and expose a workload

kubectl create deployment web --image=nginx --replicas=3
kubectl expose deployment web --port=80 --type=LoadBalancer
kubectl get svc web --watch

Google Cloud provisions an external load balancer.

6. Enable node autoscaling

Standard clusters use the flags above; Autopilot scales automatically. Combine with an HPA for end-to-end elasticity.

Verification

Confirm nodes are Ready and pods Running. Browse to the load balancer IP and confirm the app responds. In the Cloud Console, confirm the cluster and node pool report healthy.

Next Steps

Enable Workload Identity so pods access Google Cloud APIs securely, use GKE Ingress or the Gateway API for HTTP routing, and store images in Artifact Registry. Codify the cluster in Terraform, and delete demo clusters with gcloud container clusters delete demo --region us-central1 to avoid charges.