Algorithm for Containers’ Persistent Volumes Auto-scaling in Kubernetes

The article describes a proposal of an algorithm for Persistent Volumes auto-scaling in a container orchestration system - Kubernetes. Auto-scaling stands for decreasing/increasing target resources depending on some load. In case of horizontal Volumes scaling, we vary the number of Volumes, while in the vertical case we vary the size of each Volume. Also we introduce mixed scaling that includes both: horizontal and vertical as sub steps to reach a desired state. It is proposed to choose from the methods above based on free storage capacity of Kubernetes cluster and its nodes. Approaches described in the article allow to use persistent storage resources in an efficient way without manual interaction with Kubernetes cluster.