Stateful versus stateless workloads

Kubernetes is designed for both stateful and stateless applications. To maintain stateless workloads in Kubernetes, we can freely delete and replace containers without any additional concerns. The stateful application usually has storage attached either locally or in a remote location, as it needs to hold client data. That data could be short-lived or non-persistent storage, which means that it is just maintained until the expiration of a session. An example of this is the Redis cache on Kubernetes. Another use case is when the data needs to be held for long enough by using persistent storage so that it can be used on-demand. An example of the latter is the MongoDB operator for Kubernetes. The whole story is much more complicated than it seems but it all starts with Kubernetes volumes.

Kubernetes volumes represent the concept of storage in Kubernetes. As mentioned in Chapter 1, Kubernetes Overview, the volumes in Kubernetes are managed by storage...