# Google Kubernetes Engine and App Engine — Google Cloud Platform

Source: https://www.skillbyai.com/en/gcp/c-gke

> Decide when you need Kubernetes and which GKE mode to use.

## Managed Kubernetes in two modes

**Google Kubernetes Engine (GKE)** is managed Kubernetes; Google runs the control plane and handles upgrades through **release channels**. **Autopilot** mode also manages nodes: you deploy pods and pay for the CPU, memory and storage they request, with security hardening applied by default. **Standard** mode gives you control of node pools, machine types and node configuration, and you pay for the nodes. Use GKE when you need the Kubernetes API itself: many services with complex networking, custom operators, stateful workloads, GPUs or portability of Kubernetes manifests and Helm charts. **App Engine** is Google's older platform-as-a-service for web apps; it still runs, but new projects usually choose Cloud Run. Store images in **Artifact Registry**, the successor to Container Registry.

## An Autopilot cluster and a deployment

Autopilot sizes nodes from pod requests, so set them deliberately.

```bash
gcloud container clusters create-auto shop-cluster --region=asia-south1
gcloud container clusters get-credentials shop-cluster --region=asia-south1

kubectl create deployment orders --image=asia-south1-docker.pkg.dev/shop-dev-123456/apps/orders:1.0
kubectl set resources deployment orders --requests=cpu=250m,memory=512Mi
kubectl expose deployment orders --port=80 --target-port=8080 --type=LoadBalancer
```

## Cloud Run first, GKE when needed

Many teams adopt Kubernetes for a handful of stateless HTTP services and spend their time running clusters. Start with Cloud Run; move to GKE when you hit a concrete need it cannot meet.

**Quiz:** In GKE Autopilot, what are you primarily billed for?

- [ ] The control plane only
- [ ] Each kubectl command
- [x] The resources your pods request
- [ ] Node pools you create manually

*Answer:* The resources your pods request. Autopilot manages nodes and bills by pod resource requests.