I have been out of this for over a decade but the plan IBM was selling was redundant sites. You would have a mirror site 1000 miles away that could handle all of your critical applications with a moment's notice. That way you could switch over in the few minutes a basic UPS gives you and avoid all that redundancy on site. Money wise it really may end up cheaper, or at least that was what the selling point was. With national broad band capability, it is not really important where you are. I know IBM maintained mirrors in several places for our internal operations as far back as the 70s so they had experience.
The redundant site might actually be backing up dozens of sites but as long as they are not all down at once a minimal amount of hardware will do. Even if there is a bottleneck, slow is better than nothing at all.
