For each new replication that you configure, an initial full synchronization operation is performed. During this operation, vSphere Replication copies the whole data from the source virtual machine to a placeholder vApp on the target site.
Due to VM size or network bandwidth, the initial full sync might take a long time. Therefore, you might choose to copy the source VM to the target site by using removable media, or other means of data transfer. Then you can configure a replication and use the VM copy on the target site as a replication seed. When a replication is configured to use a seed vApp, vSphere Replication does not copy the whole source VM to the target site. Instead, it copies to the seed vApp only the different blocks between the source VM and the seed.
vSphere Replication stores the replication data in the seed vApp. No copies of the seed vApp are created. Therefore, a seed vApp can be used for only one replication.
Creating Seed vApps in the Cloud
Seed vApps on the target site can be created in the following ways.
Offline data transfer: You can export a VM as an OVF package and let a Cloud service administrator import the package in your cloud organization.
Clone a VM: A VM in the org virtual data center can be cloned to create a seed vApp. vSphere Replication calculates checksum and exchanges the different blocks from the replication source to the seed vApp.
Copy over the network: A source VM can be copied to the cloud organization by using means other than vSphere Replication to transfer source data to the target site.
The size and number of disks, and their assignment to disk controllers and bus nodes must match between the replication source and the seed VM. For example, if the replication source VM has two disks of 2 GB each, one of them assigned to SCSI controller 0 at bus number 0, and the second one assigned to SCSI controller 1 at bus number 2, the seed vApp that you use must have the same hardware configuration - 2 disks of 2 GB each, at SCSI 0:0 and at SCSI 1:2.