To recover a workload to a previous state, you can use point-in-time or stored instances. To avoid the automatic retention of point-in-time instances, you can store particular point-in-time instances. The stored instances do not change and you can use them to recover the workload to the stored instance, regardless of the overall retention period of the point-in-time instances.

VMware Cloud Director Availability supports two types of instances to which any workload can be recovered:

Point-in-time instances
The point-in-time instances are retained and automatically rotated during the lifespan of the replication.

VMware Cloud Director Availability automatically retains a configurable number of the last point-in-time instances and allows the workload to be recovered to any one of them.

Stored instances
The automatic retention does not affect the stored instances.

After manually storing an instance, the stored instance remains unchanged and if the replication is still active, the workload can be recovered to that stored instance.

Any replication can have both stored instances and point-in-time instances, depending on the number of allowed stored and point-in-time instances by the policy.

The automatic retention of a point-in-time instance can be bypassed by storing it. VMware Cloud Director Availability retains the stored instance until it is no longer marked as stored or until it is manually deleted. Any stored instance, without the latest one can be deleted.

Note: After a test failover, the replication can have more stored instances than the policy allows for. Performing a test failover stores the current instance and stores all its parent instances, up to the base disk. Those stored instances no longer participate in the retention policy. After a test failover, the automatically created point-in-time instances continue to participate in the retention policy. After performing a test cleanup, the instances stored by the test failover are no longer stored and again start participating in the retention policy.