Microsoft Failover Clustering has improved a few limitations with tradional MSCS which will help provide disaster recovery by geographically dispersing the nodes. Failover Clustering removed the restriction of requireing the nodes on the same subnet so at least they can be stretched and also eliminated the 500ms heartbeat requirement which should allow them to be stretched further and account for IP latency. However, there is still the same limitation and potential single point of failuver by having the shared storage requirement but there are solutions out there that allow you to get around this to stretch the clusters beyond their limits and provide dedicated storage for each node. You will still need to be consciense of bandwidth and the amount of data that needs to be synched between the nodes but that is all a part of the planning process. I think you will start to see more people stretch the nodes of this solution architected above between offices either near or far and many are doing it today for their business critical systems and have included that architecture into their business continuity plans.