Failback is the process where protected VMs and data are returned from the recovery site back to the protected site. Operation is then shifted back to the protected site.
Failback can be full (restoring the service and compute clusters) or partial (only restoring compute clusters). Moving protected VMs to a different datacenter is possible and handled in a similar fashion. Other failback modes of operation can also be performed to support custom use case configurations. They are described in separate documentation.