If you use the same snapshot for repeated failback or restore operations, it can trigger a longer, non-incremental restore.

This issue occurs with both high-frequency snapshots and standard-frequency snapshots. A full restore transfers the entirety of the VM contents from cloud snapshots, while an incremental restore only transfers the blocks which are different from the target snapshot.

Incremental restore is more rapid, and preferred where possible, but some scenarios make a full restore required.

Standard-frequency Snapshots

For users whose protected site is a VMware Cloud on AWS SDDC with a protection group using standard-frequency snapshots, this issue can occur due to the loss of change tracking information during a failback or restore operation. This issue does not affect on-premises protected sites.

If your protected site is an VMware Cloud on AWS SDDC and you have performed a prior failover-failback for a protection group leveraging standard frequency snapshots, then any subsequent restores to a snapshot taken before the prior failback requires a full, non-incremental restore.

For example, in the following scenario: 
snapshot 1

snapshot 2

[…]

snapshot n

failover-failback to snap n

Any subsequent restores to snapshots 1-n after this point require a full restore.

Workaround:

Only perform failback with snapshots taken after previous VM restore operations.

High-frequency Snapshots

This issue can occur when using high-frequency snapshots on any protected site type (on-premises or an SDDC).

If you are using high-frequency snapshots in a protection group, performing multiple consecutive VM restore operations (failback or single VM restore) performed without intervening snapshots results in a full, non-incremental restore.

Workaround:

After performing a failback or VM recovery, take a snapshot before performing failback operations are attempted. This snapshot does not have to be used for failback or restore operations.

For example, the following scenario would result in a full restore: 
snapshot 1

failover-failback to snap 1

failover-failback to snap 1
This procedure would restore the VM incrementally: 
snapshot 1

failover-failback to snap 1

snapshot 2

failover-failback to snap 1