Demo Applications

Keda CPU Scaler Example

Context

This demo application shows how to scale the Kubernetes workloads using KEDA API based on a simple CPU consumption case.

See the KEDA documentation:

• API of the ScaledObject custom resource
• Available scalers

The KEDA demo application consists of:

• order-service deployment (serving as a scale target)
• busybox deployment (generating trafic)
• scaled object using a simple CPU-based trigger

Procedure

1. Deploy the demo application:

kubectl apply -f examples/keda-cpu-scaler-demo.yml

2. Verify the successful deployment of the demo application:

You should see that scaled object is created and has a status READY:

kubectl get scaledobjects.keda.sh -n keda-demo
NAME                        SCALETARGETKIND      SCALETARGETNAME   MIN   MAX   TRIGGERS   AUTHENTICATION   READY   ACTIVE   FALLBACK   AGE
orders-service-cpu-scaler   apps/v1.Deployment   orders-service    1     10    cpu                         True    True     Unknown    8m3s

You should also see that after a while, KEDA has engaged the Kubernetes HorizontalPodAutoscaler, which controls the number of replicas of the target deployment.

kubectl get hpa -n keda-demo
NAMESPACE   NAME                                 REFERENCE                   TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
keda-demo   keda-hpa-orders-service-cpu-scaler   Deployment/orders-service   80%/30%   1         10        4          31s

Keda Prometheus Scaler Example

To experience how Kyma's Keda module can complement other Kyma components, read Scale to Zero With Keda.

Dynatrace Request Rate Scaler Example

To learn how to set up the infrastructure needed for request rate-based autoscaling using Dynatrace and the Keda and Istio modules in the Kyma environment, see Dynatrace Request Rate Scaler Example.