Tutorial - Scale applications in Azure Kubernetes Service (AKS)

If you followed the previous tutorials, you have a working Kubernetes cluster and Azure Store Front app.

In this tutorial, part six of seven, you scale out the pods in the app, try pod autoscaling, and scale the number of Azure VM nodes to change the cluster's capacity for hosting workloads. You learn how to:

Before you begin

In previous tutorials, you packaged an application into a container image, uploaded the image to Azure Container Registry, created an AKS cluster, deployed an application, and used Azure Service Bus to redeploy an updated application. If you haven't completed these steps and want to follow along, start with Tutorial 1 - Prepare application for AKS.

Manually scale pods

1. View the pods in your cluster using the kubectl get command.

   kubectl get pods
   

   The following example output shows the pods running the Azure Store Front app:

   NAME                               READY     STATUS     RESTARTS   AGE
   order-service-848767080-tf34m      1/1       Running    0          31m
   product-service-4019737227-2q2qz   1/1       Running    0          31m
   store-front-2606967446-2q2qz       1/1       Running    0          31m
   

2. Manually change the number of pods in the store-front deployment using the kubectl scale command.

   kubectl scale --replicas=5 deployment.apps/store-front
   

3. Verify the additional pods were created using the kubectl get pods command.

   kubectl get pods
   

   The following example output shows the additional pods running the Azure Store Front app:

                                     READY     STATUS    RESTARTS   AGE
   store-front-2606967446-2q2qzc     1/1       Running   0          15m
   store-front-3309479140-2hfh0      1/1       Running   0          3m
   store-front-3309479140-bzt05      1/1       Running   0          3m
   store-front-3309479140-fvcvm      1/1       Running   0          3m
   store-front-3309479140-hrbf2      1/1       Running   0          15m
   store-front-3309479140-qphz8      1/1       Running   0          3m
   

Autoscale pods

To use the horizontal pod autoscaler, all containers must have defined CPU requests and limits, and pods must have specified requests. In the aks-store-quickstart deployment, the front-end container requests 1m CPU with a limit of 1000m CPU.

These resource requests and limits are defined for each container, as shown in the following condensed example YAML:

...
  containers:
  - name: store-front
    image: ghcr.io/azure-samples/aks-store-demo/store-front:latest
    ports:
    - containerPort: 8080
      name: store-front
...
    resources:
      requests:
        cpu: 1m
...
      limits:
        cpu: 1000m
...

Autoscale pods using a manifest file

1. Create a manifest file to define the autoscaler behavior and resource limits, as shown in the following condensed example manifest file aks-store-quickstart-hpa.yaml:

   apiVersion: autoscaling/v1
   kind: HorizontalPodAutoscaler
   metadata:
     name: store-front-hpa
   spec:
     maxReplicas: 10 # define max replica count
     minReplicas: 3  # define min replica count
     scaleTargetRef:
       apiVersion: apps/v1
       kind: Deployment
       name: store-front
       metrics: 50 # target CPU utilization
   

2. Apply the autoscaler manifest file using the kubectl apply command.

   kubectl apply -f aks-store-quickstart-hpa.yaml
   

3. Check the status of the autoscaler using the kubectl get hpa command.

   kubectl get hpa
   

   After a few minutes, with minimal load on the Azure Store Front app, the number of pod replicas decreases to three. You can use kubectl get pods again to see the unneeded pods being removed.

Manually scale AKS nodes

If you created your Kubernetes cluster using the commands in the previous tutorials, your cluster has two nodes. If you want to increase or decrease this amount, you can manually adjust the number of nodes.

The following example increases the number of nodes to three in the Kubernetes cluster named myAKSCluster. The command takes a couple of minutes to complete.

You can also autoscale the nodes in your cluster. For more information, see Use the cluster autoscaler with node pools.

Next steps

In this tutorial, you used different scaling features in your Kubernetes cluster. You learned how to:

In the next tutorial, you learn how to upgrade Kubernetes in your AKS cluster.