Hands-On Microservices with Kubernetes

Abstracting storage

Kubernetes has an abstract storage model, based on volumes and persistent volume claims. It supports a large number of storage solution in-trees. This means those storage solutions had to be built into the Kubernetes code base.

Early on (in Kubernetes 1.2), the Kubernetes team introduced a special type of plugin called FlexVolume that provided an interface for out-of-tree plugins. Storage providers could provide their own drivers that implement the FlexVolume interface and could serve as a storage layer without modifying Kubernetes itself. But, the FlexVolume approach was still pretty clunky. It required installing special drivers on each node and, in some cases, on the master too.

In Kubernetes 1.13, the Container Storage ...