Having completed site selections for many data centers and in about 20+ countries, including completing site selections for Yahoo, Google, Facebook, Equinix, Exodus, and many others, I’ve learned quite a few things. I’ve been part of the changing criteria evolving from just being near fiber lines to adding in power capacity, energy price, sales taxes, property taxes, and now including climate, carbon impact and water supply as well. I think we’ll soon see income tax added in to as a major cost driver for site selections. Having been doing this for over a decade, I’ve taken on these new elements of data center site selection, driving the focus on them. Nearly 10 years ago I considered power capacity, energy price, water supply, carbon intensity of power supply, climate and taxes, only to see the industry finally accepting all of these principals as primary decision factors.
While risk of natural and human disasters has always been a part of every data center site selection, it has seriously changed. My 20+ page checklist of hundreds of items from nearby man holes covers, flight paths and train tracks to nearest police station has not been as heavily used, as it seems to add less value than thinking about the BIG natural disasters that can occur and unforeseen human caused disasters. While we used to worry about trucks of guys jumping out with AK47’s to break into a data center, the reality is, this is a thin probability and one that is difficult to prevent. Meanwhile, the ones we can prevent that are known, potential and unforeseen, are the ones we have not focused on well.
For example, has anyone thought about their utility system being hacked and shut down for an extended period of time? Have you asked your electric utility if they are NERC CIP compliant to ensure that they have a much lower chance of being hacked and shut down? Have you thought about your electric utility meter, water meter, main switchboard and generator switchgear being connected to the Internet and/or your utilities and thus being able to be hacked into, shut down, or damaged?
And the main thing, how about natural disasters? As an industry, we’ve built data centers in seismically active areas (i.e. Japan, California, Oregon (also with extreme tsunami risks) and Washington) and build so the building stays up but don’t think about all of the IT gear shuffling about and the personnel getting hurt. A building that stands while the IT gear is rolling around like marbles isn’t a data center that will sustain an earthquake, only one that will memorialize that happened while we rebuild the inside.
We build data centers in hurricane and tornado areas (Texas, Kansas, Nebraska, North Carolina, South Carolina, Virginia, Georgia) and build for them pretty well, but do we think about what has not to come yet but likely will?
I’ve written before about the most dangerous seismic area in the US not being on the West Coast or even the East Coast, but yet the Madras Fault being right under Kansas City, St. Louis and a large part of middle America.
Lately we’ve had tremendous flooding along the Mississippi and Missouri Rivers; yesterday tremendous dust storms hit Arizona–look at these amazing photos of a dust storm that hit Phoenix yesterday. (http://www.time.com/time/photogallery/0,29307,2081646_2290849,00.html). And likely future heat storms will add to the dust storms in the Phoenix area. Do you want your data center operating in this?
Severe hurricane and tornado frequency has increased many fold over the last decade and we saw more serious renditions of each over the last several months, including in places we weren’t expecting to see them, such as Massachusetts and Missouri, where tornados tore thru very robust buildings, even a hospital data center (http://www.datacenterdynamics.com/focus/archive/2011/06/missouri-tornado-destroys-hospital-data-center). Look at these photos of devasatation in Alabama from recent tornadoes: (http://www.nytimes.com/2011/05/05/us/05missing.html?_r=1&nl=todaysheadlines&emc=tha23). Imagine you and your employees following something like this–would they even come to work? One would likely need to shut down their data center for human resources even if everything kept working.
My point is not to overlook the seriousness of your data center site selection. Consider what MAY happen, with some probability, and don’t assume that just because it hasn’t happened, that it really hasn’t or won’t. Research the probabilities. The web is wonderful tool for this information, and so are your data center site selection experts at MegaWatt Consulting and others. Use us to help you avoid future problems.
Stay healthy and let’s help each other grow our industry. KC Mares