I've been reading all the best practice papers and articles I can find on data center moves. Everything I read on planning, testing and moving is supposed to lower risk. From what I can gather, the definition of lower is - 80% failure in some major way (thankfully these folks don’t make airplanes).
These plans are like the old Rube Goldberg machines; make elaborate plans, test elaborate plans, pat each other on the back about elaborate planning, then put plan into action and immediately engage the part of that which deals with the catastrophe that is occurring. This has an elegant insanity wafting around it like waste water treatment plant on a warm summer day. In short, fork-lifting my fragile data center equipment into a truck, for a ride across crater pocked roads while dodging scooters and octogenarians in land-yachts, is insane.
The only plan I’ve seen that makes any sense is Swing Gear. Why isn’t this the only way data centers are moved? Renting all or at least a good part of the equipment to replicate the environment, testing, and stabilizing the new center just makes sense. In the past this has been too expensive, time consuming or impossible to find enterprise class equipment in one place, but the emergence of mega-resellers willing to rent equipment for short-term migrations, augmentation or testing has changed the landscape dramatically.
Canvas Systems supports Swing Gear moves and migrations with serial number for serial number replication. This mitigates almost all the risk and makes testing and troubleshooting easy. Here's a link to the site http://www.canvassystems.com/Services/Server-Rental/Data-Center-Migration.aspx.