1) Be cognizant of all hazards that could affect your data center. There’s a standard list that includes proximity to airports, rail, major highways, meteorological/natural hazards, environmental hazards (including hazardous waste or chemicals), flood plains and seismically active areas among 68 other considerations.
2) Make sure that the utility serving the area can provide adequate amounts of water and power. With regard to the latter, the quality of the power and dependability of the source are also extremely important. Finally, pricing can make or break the viability of your project. Every one cent per kWh difference with regard to usage of 1MW of power equates to $250,000 per year in either excess cost or savings.
3) Make sure that physical structure of the building supports the data center use. In most cases, it’s more costly to convert an existing building to a data center use than to demolish the existing building and construct a purpose-built data center. I’ve seen a number of data centers in subterranean space. These spaces are subject to water table levels, water incursion from spaces above, and potential seismic threats. So, ground level, single-story structures are typically the preferred.
4) It’s best to be located in a building that has no mixed use. That is, the building is occupied only by your company, by other data center operators, or enterprises operating a data center. Having retail, office, or residential use in the same building is usually not prudent.
5) Connectivity. Though we’ve come a long way since 10 or 15 years ago, the availability of abundant connectivity is still a serious requirement. As a matter of fact, a number of operators are using network redundancy to avoid building multiple high tier-rated data centers (see items number 14 below).
6) Disaster recovery, availability and redundancy. Locating a disaster recovery, or secondary data center too close to a primary data center just isn’t a good idea. Some enterprises consider the threshold distance as anything closer than 500 miles and that may even be too close depending on other factors.
7) Plan for today’s business needs, but build in flexibility to adapt to changes. Build what you need now, but make sure that in the future, the back-of-house infrastructure equipment and design can accommodate growth and changes. Of course, no one has a crystal ball, so it’s impossible to know what the latest-greatest, lowest-PUE, highest-sustainable design will be, but modularity and flexibility will support your transition.
8) And while we’re on the topic of sustainability, even though it doesn’t seem like anyone wants to pay extra for being green, there may be more interest generated for this in the future if cap-and-trade is enacted, globally. Moreover, going green means not only being responsible, but also operating more efficiently, reducing energy usage, increasing cooling efficiencies and reaping the monetary savings that these practices create.
9) Be realistic about pricing. The cheapest bid is probably not going to deliver the resilient, available data center that you’ve envisioned. That said, do your homework. Hire a proficient project manager to act as your representative and bid the job with multiple designers, contractors and vendors. The competitive landscape will create negotiating leverage.
10) Manage and measure your data. Research DCIM programs and learn which will serve your purposes and your data center. Use the information to drive better efficiencies and reduce your costs. That way you can buy more IT gear.
11) And while we’re talking about IT gear…Virtualize wherever possible. Most enterprises today are barely virtualizing more than 20% of their servers. Not all applications can run in a virtual environment, but creating opportunities will drive down your operating costs significantly.
12) Build in security on all levels. Physical security and digital security are the key to complying with auditing standards under which your company or your clients are operating. Compliance procedures, firewalls and other security measures are essential to uptime and business interruption planning.
13) Your business continuity plan not only includes having a disaster recovery site, but also constantly testing the equipment infrastructure at each site to ensure that when it’s called upon to support an outage your operations function seamlessly.
14) And while we’re speaking about business continuity planning, the most prevalent thought leadership that I’m hearing is that having multiple data centers that are N+1 in geographically diverse areas may provide much more redundancy that having one 2N+1, concurrently maintainable data center.
Finally, Business Case trumps almost everything else! What’s your raison d’être? Why do you need a location in that particular place, or is there somewhere else that is better? If you’re a colo provider, can you sustain a business in the target location? If you’re an enterprise user, will the location support your use with a high-quality labor pool? Are there business and tax incentives to offset development costs? What makes the location unique? And once you understand your business case, you may have to overlook or modify any of the 14 items above to satisfy your business case parameters.
I hope this provides some guidance and insight into a few issues that are typical drivers for most decision-makers in the data center sector.