- Before invoking any APIs from PowerProtect Data Manager to vCenter, PowerProtect Data Manager takes care of the configuration of the VM. For example, if a disk should be added or removed because the VM changed after the snapshot was taken, the disk might be added. For this reason, the reconfiguration part is performed beforehand.
- A metadata only snapshot is taken using the same snapshot creation flow described in the section Transparent snapshots creation.
- PowerProtect Data Manager invokes the restore operation, directing which VM should be restored to what point in time.
- The VM is checked to determine if it is in a powered off state. If it is not, the VM is powered off.
- PowerProtect Data Manager locates the relevant ESXi host through vCenter, and the ESXi host communicates with the TSDM to initiate a restore operation.
- For a restore workflow, first, reserve all the areas of the disks of the VM that should be reverted to the previous point in time, for the following reasons:
- To minimize the data you transfer from the protection storage back to the disk.
- To identify which parts of each disk have changed since the point in time to which we are trying to revert.
- To resolve the changes (described above), the restore workflow leverages two Get Diff APIs, namely Get VAIO Diff API and the DD Get Diff API.
- The ESXi host resolves the Get VAIO Diff API. This difference includes the changes that the VM has made to the disks since the last point in time that was previously synced or backed up to the PowerProtect Appliance. The VAIO Diff uses the metadata only snapshot, taken before the VM was powered off, to get the details on what was never written to the PowerProtect Appliance.
- The DD Get Diff API provides the delta details between the last sync point-in-time and the one to revert. It also merges the delta with the delta returned from the previous step. This step provides the complete set of extents that can now be read from the PowerProtect Appliance.
- From the point-in-time copy to which the user wants to revert, data is read and then finally written on top of the VMDKs.
- After all the data movement is complete, the TSDM sends an acknowledgment to the ESXi host that the restore process has been completed.
- vCenter marks the task completed and sends an acknowledgment to PowerProtect Data Manager for catalog update.
- The VM can now be powered on and should have been successfully reverted to the previous point in time.
- The metadata only snapshot can now be retired, in the same manner as after every delta sync operation.
Note the following:
- Multiple streams are opened on the TSDM (when ESXi receives the request) to achieve restore parallelism. You can achieve a higher level of parallelism if the VMs are spread across multiple ESXi hosts.
- Restore in parallel supports up to eight disks of a VM using transparent snapshots.
- Only Restore to Original and New are supported with PowerProtect Data Manager 19.9 release using transparent snapshots. Although instant access and File Level Restore are not supported at this time with transparent snapshots, they would default to a restore workflow using VM Proxies (vProxy) if performed.
Figure 15. Restoring virtual machines