Pod Autoscaling
Autoscaling is a widely accepted practice in cloud computing which entails automatically adjusting the amount of computational resources based on load.
Autoscaling automates resource management and as such has a significant impact on cluster costs.
Well configured autoscaling can make your cluster lean, efficient and reliable.
Badly configured autoscaling can generate waste and reduce availability.
Kubernetes comes with add-ons that allow us to autoscale pods in 2 distinct ways:
-
Horizontally
By changing the number of pod replicas in a workload (Deployment, StatefulSet)
-
Vertically
By changing the amount of resources (CPU, memory) allocated to each individual pod in a workload.
Horizontal Pod Autoscaling
Horizontal pod autoscaling is enabled by the following add-ons:
(Follow the links for autoscaling optimization recommendations)
Vertical Pod Autoscaling
Horizontal pod autosscaling is enabled by the following add-ons:
(Follow the links for autoscaling optimization recommendations)
