Forum Moderators: phranque
What I know is (and the terminology alone may help in your research) that you're talking about "mirror site fail over". As a rough sketch, this involves two main functions: monitoring the primary data center's health, and seamlessly redirecting traffic to the mirror if signs of trouble arise.
An added feature can be "load balancing" which doesn't wait for a failure, but instead distributes traffic between the data centers on a regular basis and continually monitors factors like latency and packet failures.
There are some major companies that offer outsourced solutions for this kind of thing. Being faint of heart myself when it comes to this league of heavy duty technology, I know I would explore that kind of service first.