[Neoeclectic]

You could write a book about this topic so offering a thorough explanation would be tough.

Generally, most serious data center business will have disaster recovery plans. A disaster recovery plan pretty much covers all disasters that could bring a data center to its knees such as earthquakes, hurricanes, lightning strikes, torrential rain, meteor strike, and etc. So as a high level explanation there should be a plan in place that details the follow up procedures in the event of a disaster.

Most commonly redundant data center sites, setup in separate and distinct disaster zones, is a commonly used practice. In other words, you wouldn't have your primary data center and your backup data center down the street from one another, or even within the same state. Serious data center planners always situate the two data centers so that no two share the same central power grid, seismic zone, and even network backbones. You can think of it as N+1 redundancy inwhich both data centers are completely independant from one another in all respects. So if an earthquake takes your primary data center out that same earthquake won't touch your backup site. It's common practice for larger companies to build a data center on the east cost then build a second one in Chicago, Houston, or even as far away as Washington state.

There are a multitude of precautions taken within the data center itself as well to protect against natural disasters. Some data centers like to anchor their cabinets to the foundation to resist earthquakes. Some will add lightning protection systems to protect against lightning strikes, and on, and on, and on. This could be discussed into ad nauseum. The best thing to do is do a little research on disaster recovery and disaster prevention because there are a lot of subsets of topics that has to be addressed and covered.