I am using the VNXe3200 to serve up NFS datastores to my 3 ESXi 5.5 hosts.

In the document VNXe3200 High Availability - A detailed review it shows setting up SP A and SP B on different subnets in the NFS configuration details.  My question is why?  How does using different subnets help HA as opposed to two different IP's on the same subnet?  I currently have eth2 and eth3 aggragated using LACP and two different LAG groups in the switch, but both on the same subnet.  Should I rethink this config?

Thanks for your help.