An administrator may encounter VMs that need to be replicated, but it might be easier for them to be first copied over to the target site. In this situation, the relevant volume can be copied to external media, brought to the replication site, added to the remote datastore, and selected as a "Preseed" of the replicating volume. Alternatively, a relevant pre-existing volume on a datastore at the replication site could also be used. Instead of a full initial sync, a delta-sync occurs at this point to compare the volumes for any changes that may have occurred between the initial copy and the selection of Preseeding.
It may also be desirable to preseed in the event that a VPG needs to be recreated. This circumvents the need to perform an initial sync again, when there is already a previously-used recovery volume at the replication site.
A disk or disks might be too large to be sent over WAN or the customer may wish to shorten the amount of time the resulting sync might take.
If preseeding is being used in a manner in which a previous recovery volume needs to be located -- for example, when recreating a VPG -- it is important to understand the directory structure of recovery volumes as described in Recovery Volume Directory Structure.
In order to configure preseeding, follow these steps:
- For a previously configured VPG, click the pencil icon in the "VPG Name" column. For a new VPG, follow the next steps after adding VMs
- Click on the "Storage" tab. Check the VM you wish to preseed, and click "Edit Selected"
- For Volume Source, set the drop-down to "Preseeded volume"
- Select the datastore where the volume is located at the recovery (target) site
- Click on the ellipsis (...) in the "Path" field to browse for the volume you wish to preseed. Select the volume
- Click OK on the "Edit Volumes" window
- Click "Next" to continue configuring the VPG or "Done" if complete with configuration
- The preseeded volume will now enter delta sync as it is compared with the production volume
In some instances, copying the protected virtual disk(s) to the recovery site will take substantial time if the disk is particularly large. In this case, we recommend using the preseed option so that the initial synchronization of the VPG upon creation will be faster since a delta-sync will synchronize any changes made to the virtual disk rather than the entire disk. Note that when not using a preseeded disk, the initial synchronization must copy the entire disk over the WAN. In such a scenario where a VMDK is not available for preseed at the recovery site, an administrator will need to manually place copies of the source VM's VMDKs at the recovery site in order to utilize the option to preseed.
To manually copy VMDKs for preseeding, follow the procedure below:
- Clone the source VMs you wish to use for preseed prior to creating a new VPG in Zerto. Follow the steps outlined in the following VMware KB to create the VM clone: https://pubs.vmware.com/vsphere-50/index.jsp?topic=%2Fcom.vmware.vsphere.vm_admin.doc_50%2FGUID-5C504B67-CDB3-42FC-8B3B-737201A725DD.html
NOTE: You cannot create a clone from a virtual machine that is powered on or suspended. You must first power off a virtual machine before you can make a clone
- After the VM clone task(s) have completed, check if any of the cloned VMs are powered on. Power off any cloned VMs if they were powered on after creation
- In vSphere, right click on the cloned VMs and select "Edit Settings." From "Edit Settings," select each hard disk and click "Remove." Ensure that the Removal Option "Remove from Virtual Machine" is selected prior to clicking OK to commit the changes
NOTE: If the hard disk removal option "Remove from virtual machine and delete files from disk" is selected, the underlying VMDKs will be deleted
- After the hard disks have been removed from the cloned VMs, copy the VM clone VMDKs to an external hard drive/array for sneaker-net transfer to the recovery (target) vSphere site datastore or transfer the files directly to a datastore on the recovery (target) vSphere site via NAS
- Remove the VM clones from inventory in vSphere by right clicking on the VM and selecting "Remove from Inventory." The virtual disks (.VMDKs) for each VM will remain in the cloned VM directory on the specified datastore/hard disk
NOTE: Only the virtual disks (.VMDK) files need to be moved to the specified datastore. If the folder containing the disks is moved, vCenter's default behavior is to inflate thin provisioned disks within the directory which may not be desirable. By moving only the VMDKs, the existing provisioning method for the disk will be preserved
- Power on the source VMs that were previously cloned
- In the Zerto GUI, select "New VPG" and proceed with the necessary VPG creation steps. In the Zerto Virtual Replication 4.0 versions, preseeding is specified in the "Storage" tab. Follow the preseed steps outlined above in order to preseed the volumes that were manually copied to the recovery (target) site
Note the following:
- IDE disks cannot be selected as preseeded disks.
- Only disks with the same size as the protected disks can be selected when browsing for a preseeded disk.
- If the preseeded disk is greater than 1TB on NFS storage, and the target ESXi version is 4.0 or 4.1, the VPG creation might fail. See VMware's Block size limitations of a VMFS datastore KB article for more information.
- If the protected disk is a RDM disk, it can be used to preseed to a recovery VMDK. Zerto Virtual Replication ensures that the VMDK disk size is a correct match for the RDM disk.
- If the VPG is being defined for a Zerto Organization (ZORG), the location of the preseeded disk must be defined in the Zerto Cloud Manager. For details, see the Zerto Cloud Manager Administration Guide
- Preseeding to Amazon Web Services (AWS) or Microsoft Azure is not supported