skip to content
Astro Cactus

Autoscaling in Kubernetes

/ 3 min read

kubernetes , k8s , autoscaling , hpa , vpa , devops , cloud-native , containers , series:kubernetes:15

Understanding Kubernetes Autoscaling

Kubernetes provides three types of autoscaling:

  1. Horizontal Pod Autoscaling (HPA)
  2. Vertical Pod Autoscaling (VPA)
  3. Cluster Autoscaling (CA)

Horizontal Pod Autoscaling (HPA)

Basic HPA Configuration

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: nginx-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: nginx
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50

Advanced HPA Configurations

Multiple Metrics

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: multi-metric-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: app
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 70
  - type: Pods
    pods:
      metric:
        name: packets-per-second
      target:
        type: AverageValue
        averageValue: 1k

Custom Metrics

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: custom-metric-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: app
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Object
    object:
      metric:
        name: requests-per-second
      describedObject:
        apiVersion: networking.k8s.io/v1
        kind: Ingress
        name: main-ingress
      target:
        type: Value
        value: 100

Vertical Pod Autoscaling (VPA)

VPA Installation

apiVersion: apps/v1
kind: Deployment
metadata:
  name: metrics-server
  namespace: kube-system
spec:
  selector:
    matchLabels:
      k8s-app: metrics-server
  template:
    metadata:
      labels:
        k8s-app: metrics-server
    spec:
      containers:
      - name: metrics-server
        image: k8s.gcr.io/metrics-server/metrics-server:v0.6.1

Basic VPA Configuration

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: my-app-vpa
spec:
  targetRef:
    apiVersion: "apps/v1"
    kind: Deployment
    name: my-app
  updatePolicy:
    updateMode: "Auto"
  resourcePolicy:
    containerPolicies:
    - containerName: '*'
      minAllowed:
        cpu: "100m"
        memory: "50Mi"
      maxAllowed:
        cpu: "1"
        memory: "500Mi"

VPA Modes

  1. Auto Mode
updatePolicy:
  updateMode: "Auto"
  1. Initial Mode
updatePolicy:
  updateMode: "Initial"
  1. Off Mode
updatePolicy:
  updateMode: "Off"

Cluster Autoscaling (CA)

AWS EKS Configuration

apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-autoscaler-config
  namespace: kube-system
data:
  config.yaml: |
    ---
    autoDiscovery:
      clusterName: my-eks-cluster
    awsRegion: us-west-2
    extraArgs:
      scale-down-delay-after-add: 10m
      scale-down-unneeded-time: 10m
      skip-nodes-with-system-pods: false

GKE Configuration

apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-autoscaler-config
  namespace: kube-system
data:
  config.yaml: |
    ---
    clusterName: my-gke-cluster
    cloudProvider: gce
    nodeGroups:
    - minSize: 1
      maxSize: 5
      name: default-pool

AKS Configuration

apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-autoscaler-config
  namespace: kube-system
data:
  config.yaml: |
    ---
    autoscalingGroups:
    - name: nodepool1
      minSize: 1
      maxSize: 5

Best Practices

1. HPA Best Practices

  • Set appropriate min/max replicas
  • Choose metrics carefully
  • Consider scaling behavior
  • Monitor scaling events

2. VPA Best Practices

  • Use appropriate update mode
  • Set reasonable min/max limits
  • Test with non-critical workloads first
  • Monitor resource recommendations

3. CA Best Practices

  • Configure scale-down delays
  • Set appropriate node group sizes
  • Monitor cluster utilization
  • Consider cost implications

Monitoring Autoscaling

Prometheus Metrics

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: autoscaling-monitor
spec:
  selector:
    matchLabels:
      app: autoscaling-metrics
  endpoints:
  - port: metrics

Grafana Dashboard

apiVersion: integreatly.org/v1alpha1
kind: GrafanaDashboard
metadata:
  name: autoscaling-dashboard
spec:
  json: |
    {
      "title": "Autoscaling Metrics",
      "panels": [
        {
          "title": "HPA Metrics",
          "type": "graph"
        }
      ]
    }

Troubleshooting

Common Issues

  1. HPA Not Scaling
Terminal window
kubectl describe hpa <hpa-name>
kubectl get --raw "/apis/metrics.k8s.io/v1beta1/pods"
  1. VPA Issues
Terminal window
kubectl describe vpa <vpa-name>
kubectl logs -n kube-system vpa-recommender
  1. CA Issues
Terminal window
kubectl logs -n kube-system cluster-autoscaler
kubectl describe nodes

Conclusion

Implementing autoscaling in Kubernetes requires careful consideration of your application’s requirements and behavior. Proper configuration of HPA, VPA, and CA ensures optimal resource utilization and cost efficiency.

Series Navigation