Hands-On Microservices with Kubernetes

Cluster autoscaling

Pod autoscaling is a gift to developers and operators – there's no need for them to manually scale services up and down or write their own half-based autoscaling scripts. Kubernetes provides a robust solution that is well-designed, well-implemented, and battle tested. However, that leaves the question of cluster capacity. If Kubernetes tries to add more pods to your cluster, but the cluster is running at maximum capacity, then the pod autoscaler will fail. On the other hand, if you over-provision your cluster just in case the pod autoscaler needs to add a few more pods, then you're wasting money. Enter the auto-scaler cluster (https://github.com/kubernetes/autoscaler/tree/master/cluster-autoscaler).

It is a Kubernetes ...