Configure KEDA

When installed, KEDA creates several custom resources. One of those resources, a ScaledObject, enables you to map an external event source to a Deployment or StatefulSet for scaling. In this lab, we'll create a ScaledObject that targets the ui Deployment and scales this workload based on the RequestCountPerTarget metric in CloudWatch.

~/environment/eks-workshop/modules/autoscaling/workloads/keda/scaledobject/scaledobject.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: ui-hpa
  namespace: ui
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: ui
  pollingInterval: 30
  cooldownPeriod: 300
  minReplicaCount: 1
  maxReplicaCount: 10
  triggers:
    - type: aws-cloudwatch
      metadata:
        namespace: AWS/ApplicationELB
        expression: SELECT COUNT(RequestCountPerTarget) FROM SCHEMA("AWS/ApplicationELB", LoadBalancer, TargetGroup) WHERE TargetGroup = '${TARGETGROUP_ID}' AND LoadBalancer = '${ALB_ID}'
        metricStat: Sum
        metricStatPeriod: "60"
        metricUnit: Count
        targetMetricValue: "100"
        minMetricValue: "0"
        awsRegion: "${AWS_REGION}"
        identityOwner: operator

A

This is the resource KEDA will scale. The name is the name of the deployment you are targeting and your ScaledObject must be in the same namespace as the Deployment

B

The minimum number of replicas that KEDA will scale the deployment to

C

The maximum number of replicas that KEDA will scale the deployment to

D

The expression uses CloudWatch Metrics Insights syntax to select your target metric. When the targetMetricValue is exceeded, KEDA will scale out the workload to support the increased load. In our case, if the RequestCountPerTarget is greater than 100, KEDA will scale the deployment.

More details on the AWS CloudWatch scaler can be found here.

First we need to gather some information about the Application Load Balancer (ALB) and Target Group that were created as part of the lab pre-requisites.

~$export ALB_ARN=$(aws elbv2 describe-load-balancers --query 'LoadBalancers[?contains(LoadBalancerName, `k8s-ui-ui`) == `true`]' | jq -r .[0].LoadBalancerArn)

~$export ALB_ID=$(aws elbv2 describe-load-balancers --query 'LoadBalancers[?contains(LoadBalancerName, `k8s-ui-ui`) == `true`]' | jq -r .[0].LoadBalancerArn | awk -F "loadbalancer/" '{print $2}')

~$export TARGETGROUP_ID=$(aws elbv2 describe-target-groups --load-balancer-arn $ALB_ARN | jq -r '.TargetGroups[0].TargetGroupArn' | awk -F ":" '{print $6}')

Now we can use those values to update the configuration of our ScaledObject and create the resource in the cluster.

~$kubectl kustomize ~/environment/eks-workshop/modules/autoscaling/workloads/keda/scaledobject \

| envsubst | kubectl apply -f-