The design of SDV was around target virtualization. If an array has 2 ports, and one is the primary, and it fails, all the hosts must be re-zoned with the back up port. Host like HP-UX and AIX would see a change in FCID which can cause time to resolve. With SDV we virtualize the primary pwwn and back up pwwn to a v-pwwn. We then zone the hosts with the v-pwwn. In the config we link the v-pwwn to the primary. If that fails, we change the link to map the p-wwn to the back up port. The hosts see no change in their target pwwn or FCID...no re-zoning is needed. You can map the v-pwwn to multiple real pwwns, but the link can only be active to one at a time. We do support SDV for initiators as well. The enhancement on that end is that if an HBA is replaced, we do not need to update the fabric zoning or the array security software. We change the link on the v-pwwn to map from the new HBA pwwn to the v-pwwn. The array sees no change in the host pwwn or FCID. Again, no zoning changes are needed.

Hope this helps,

Mike