Feature Description

Once persistent data is provisioned in a kubernetes cluster there is no easy or automated way to migrate this data to a different storage class. This feature aims to close this gap by providing a high level API that can be used to select PVCs from a given storage class and migrate them to a different storage class regardless of what type of application is using the data and whether that application is active or stopped. The migration may take a long time and should be managed by an operator from start to finish.

Why is this important

There are many reasons that a customer may want to migrate their persistent workloads to a different storage class. They may have purchased a new storage appliance and want to easily put it into service. Existing storage may be getting retired or decommissioned. Workloads that use local storage may need to be rebalanced.

Currently OpenShift lacks an easy to use feature to move data while preserving the workloads. Closing this gap will make customers more comfortable with the platform because they know that their data is agile and can move around as their storage needs change.

Acceptance Criteria

• An API that orchestrates migration of workloads based on a set of included PVCs
• Container-based workloads can be migrated when the PVCs are not in use (Pods, Deployments, ReplicaSets, StatefulSets, etc)
• VM-based workloads can be migrated regardless of whether they are running or not

Out of scope

• Live storage migration for non-VM workloads is out of scope