Adaptive scaling of Kubernetes pods

Scalability is the ability of a resource or an application to be expanded to handle the increasing demands. One of the most emphasized aspects of cloud computing is the ability of scaling up or down. Scaling can be performed by either an operator or automatically. Autoscaling algorithms are getting more and more emphasized in the field of cloud computing. In this paper we propose an adaptive autoscaler, Libra, which automatically detects the optimal resource set for a single pod, then manages the horizontal scaling process. Additionally, if the load or the underlying virtualized environment changes, Libra adapts the resource definition for the pod and adjusts the horizontal scaling process accordingly.