A Deployment manages a stateless app declaratively. It creates a ReplicaSet, which keeps the desired number of identical Pods running — recreating any that die. You change the Deployment (image, replicas) and Kubernetes reconciles.

Its headline feature is the rolling update: it gradually replaces old Pods with new ones, respecting maxSurge/maxUnavailable, so there’s no downtime. If a new version misbehaves, kubectl rollout undo rolls back to the previous ReplicaSet.

apiVersion: apps/v1
kind: Deployment
metadata: { name: web }
spec:
  replicas: 3
  selector: { matchLabels: { app: web } }
  template:
    metadata: { labels: { app: web } }
    spec:
      containers:
        - name: web
          image: myapp:1.4.2

kubectl apply -f deploy.yaml
kubectl scale deploy/web --replicas=5
kubectl set image deploy/web web=myapp:1.5.0   # triggers a rolling update
kubectl rollout status deploy/web
kubectl rollout undo deploy/web                # roll back

1. Write a Deployment with 3 replicas of an app.
2. Scale it to 5 and watch the new Pods appear.
3. Update the image and observe the rolling update, then roll it back.
4. Explain the relationship between a Deployment and a ReplicaSet.

• ↗ Kubernetes — Deployments