Note that the VRMS and VRS are the same appliance, the VRMS just has the management components enabled. Also note that traffic flows from the source host to the VR appliance (either VRMS or VRS) then to the target host datastore. It is best to have the VR appliance close to the target.
The VRMS should be located at the site with your VC (as it requires a VC to operate) and the VRS at the smaller site. This would allow for the best traffic flow during normal operation and failback (note that when you configure for failback you would select the VRMS at the larger site, during normal replication you would replicate to the VRS at the smaller site).
Also note that with VR recovery is manual and done VM by VM. If you want DR automation take a look at SRM (though note that you would need a VC at each site).
I also recommend having a look at the VR calculator to confirm that your bandwidth will be sufficient: www.vmware.com/vrcalculator