The full sync basically just reads data for the disk and checksums the data. It also makes remote requests to read the data and return a set of checksums. It would only send data if the checksums don't match. So, assuming not much of the disk has changed, there's very little except checksums going over the network. So you're going to mostly be bound by how quickly you can read the vmdk files on both sites and to a lesser degree by the the network latency between sites, and, of course, whether the link between sites remains up.
Check into whether the link to the sites is good. Are connections often getting dropped or otherwise throttled in some way? Is the latency extremely high, like hundreds of milliseconds? This can slow things down.
On the source site, make sure the storage is not overloaded with other jobs. In the case of NAS/ISCSI, also check that there aren't other network jobs hogging bandwidth. Sometimes there are periodic jobs on secondary sites, like backups, and that people often forget about that can be resource intensive.
On the destination site, do similar checks and make sure the VR server is getting a reasonable amount of Network and CPU resources.
