Use HorizontalPodAutoscaler with memory (custom) metrics / memory and CPU and explore autoscaling API versions

To execute applications, a Kubernetes cluster requires computational resources, which may need to increase or decrease based on the workload at the time or application needs. This is usually classified as “scaling”. You can run multiple instances (replicas/pods) for the application to deal with increased requirements, such as high traffic, but why not let Horizontal Pod Autoscaler (HPA) do it for you according to your customization?