Topology for Inter-datacenter VMotion

An often asked question in any of the vCloud discussions is about the probability of vMotion across data centers (over wide area network) - it is not an easy thing considering the latencies. The complications are both in terms of stretching the L2 domain between the sites and the shared storage issue across sites.
What are the business requirements for inter-datacenter vMotion?

1. Load balance/Go-green to save power: “follow the sun” or simply consolidate VMs to conserve power.
   
2. Uptimes during maintenance windows: Have a fallback plan for your applications to be available during datacenter shutdowns
   
3. Disaster Recovery/avoidance: This is an obvious reason, to fallback during DR situations

Compared to #2 and #3, #1 could be higher priority as that also gives a provision to use someone else's (like AWS) data center for peak loads (like month end report generations) instead of building up in-house data centers.

Overall, this initiative would lead to consolidation in data centers slowly and bigger players would benefit. Below is the topology of vMotion that would be demoed at VMworld, jointly by Cisco & VMware.