Your application experiences unpredictable traffic spikes. Design a comprehensive autoscaling strategy.

Complete HPA with Multiple Metrics

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: news-app-hpa
  namespace: production
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: news-app
  minReplicas: 10        # Always keep at least 10 pods (handle baseline)
  maxReplicas: 200       # Never exceed 200 pods (cost control)

  # Multiple metrics - scale based on whichever hits threshold first
  metrics:
  # Scale based on CPU utilization
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70  # Scale when avg CPU > 70%

  # Scale based on memory utilization
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 80  # Scale when avg memory > 80%

  # Scale based on custom metric (requests per second)
  - type: Pods
    pods:
      metric:
        name: http_requests_per_second
      target:
        type: AverageValue
        averageValue: "100"  # Scale when pod handles > 100 RPS

  # Scaling behavior - control how fast to scale up/down
  behavior:
    scaleUp:
      stabilizationWindowSeconds: 0  # Scale up immediately (no delay)
      policies:
      - type: Percent
        value: 100  # Double the pods at once if needed (10 → 20 → 40 → 80)
        periodSeconds: 15
      - type: Pods
        value: 20   # Or add 20 pods at once, whichever is higher
        periodSeconds: 15
      selectPolicy: Max  # Use the policy that adds more pods

    scaleDown:
      stabilizationWindowSeconds: 300  # Wait 5 min before scaling down
      policies:
      - type: Percent
        value: 50  # Remove max 50% of pods at once (slow scale-down)
        periodSeconds: 60
      - type: Pods
        value: 5   # Or remove 5 pods, whichever is lower
        periodSeconds: 60
      selectPolicy: Min  # Use the policy that removes fewer pods

How HPA Works

The HPA calculation works as follows: it checks current state (10 pods at 90% CPU), calculates target replicas as current multiplied by current utilization divided by desired utilization (10 times 90 divided by 70 equals 13 pods rounded), scales to 13 pods, then recalculates every 15 seconds based on latest metrics.

Cluster Autoscaler Configuration

# AWS EKS Cluster Autoscaler
apiVersion: apps/v1
kind: Deployment
metadata:
  name: cluster-autoscaler
  namespace: kube-system
spec:
  replicas: 1
  template:
    spec:
      serviceAccountName: cluster-autoscaler
      containers:
      - name: cluster-autoscaler
        image: k8s.gcr.io/autoscaling/cluster-autoscaler:v1.27.0
        command:
        - ./cluster-autoscaler
        - --cloud-provider=aws
        - --namespace=kube-system
        - --node-group-auto-discovery=asg:tag=k8s.io/cluster-autoscaler/enabled
        - --balance-similar-node-groups
        - --skip-nodes-with-system-pods=false
        - --scale-down-enabled=true
        - --scale-down-delay-after-add=10m
        - --scale-down-unneeded-time=10m
        - --scale-down-utilization-threshold=0.5

VPA for Right-Sizing

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: news-app-vpa
  namespace: production
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: news-app
  updatePolicy:
    updateMode: "Recommender"  # Only recommend, don't auto-apply
  resourcePolicy:
    containerPolicies:
    - containerName: news-app
      minAllowed:
        cpu: "100m"
        memory: "128Mi"
      maxAllowed:
        cpu: "4000m"
        memory: "8Gi"