It is exactly Pat's comment about "losing control of your data" that caused the U.S. Navy to set up the standby (alternate site) systems that ran the U.S. Navy Reserve personnel management database in 2005-2006 after Hurricane Katrina trashed New Orleans.
Oh, there were half-a-dozen commercial backup sites we could have used, but the problem was that with our security requirements, commercial backup sites were expensive. AND if it wasn't a true emergency, just a test run, their contract said they had the right to control the test dates to conform to their schedule.
("How expensive was it?" says the voice from the audience...) Expensive enough that to actually do a one week full-on test run once per year for the configuration we had, it was cheaper for us to buy another pair of servers for remote production and development. We bought two more Itanium server-class systems with 8 GB RAM and 4xdual CPU cores ("Longhorn" series, 8 threads each box) plus system disks, fiber-channel NAS network cards, and Gigabit Ethernet cards for less money than one test of our full configuration. AND... Congress, in their infinite wisdom, had mandated that ALL U.S. military servers would undergo remote-operation testing once per year.
