When the time comes for a company to decide on a data center strategy, there are numerous inputs that must be addressed to properly validate this decision. A wrong projection, estimate, or even a simple oversight can devastate a company, forcing financial considerations to attend to the unintended consequences of bad decisions. In other words, no amount of facility or infrastructure design can overcome the added cost of a poor site selection assessment. The process for finding the best place to locate a data center is under continual refinement to accommodate an industry that is evolutionary and constantly changing based on the concept of developing technologies.
Before anyone can build a solid data center strategy, a certain understanding of basic economics must be understood. In the data center market, conventional wisdom suggests that availability of existing data center inventory should be ample, and that single-tenant data centers improved with today's or almost to today's standards are everywhere.
This widespread perception is a fallacy. What we have in today's data center market climate is a classic economic imbalance, where demand is significantly outpacing supply. The "State of the Data Center Market," as I like to call it, is a violator of generally accepted economic principles led by a cognitive deception of what should be versus what actually is. Long story short, there's double, almost triple the demand for data center space (or more accurately gauged these days by power demand) than what the market supplies.
While the intricacies of this concept are another discussion outside of the scope of this article, it must be generally understood that data center demand is currently exceeding supply. Once this education of data center market economics has been addressed, a user can now move to build a strategy. Toward that end, let's consider the typical components that define a data center site selection strategy that most medium- to large-size data center users endure.
Generally, there are seven primary drivers that are examined when searching for the ideal place to locate a data center. In order of preference, these aforementioned drivers include environmental, power, connectivity, tax impact, incentive considerations, construction costs, and labor.
• Environmental is a function of the risk and exposure to natural disasters. This typically has the highest weighting, as issues from seismic concerns to flood plain risks can temporarily shut down or completely wipe out a data center.
• Power is usually a large driver in that the electric bill can be one of the largest ongoing monetary line items, and will completely vary in overall weight depending on the projected power demands or capacities of a data center user. When considering power, pricing stability and/or volatility is also a viable and widely accepted driving factor, which can be mitigated in many ways, e.g., determine if a market is regulated or review the generation mix of a utility in a given market.
• Both tax impact and economic incentive considerations are extremely user-specific. Generally, as a medium to large user prospecting the market, you'll want to make sure that you look at what you're doing inside the walls of the data center (revenue production) and what a community may offer given your requirements (via capital and employment forecasts) on a city, county, and state level. Currently, there are 11 states with legislated data center incentives, and an in-depth analysis of the factors mentioned above could translate into millions of dollars of savings over the life of your data center.
• Construction cost data can easily be analyzed from a site selection perspective by benchmarking against a published national index.
• And finally labor - while usually a minimal consideration given weights of these other drivers (data centers typically employ less than 100 employees) - must be observed and can easily be assessed by an examination of IT workers in an area.
Site Selection Model
So now that some typical areas of investigation for medium- to large-size data centers have been revealed, applying this market knowledge to the process is the next logical step. For simplicity, let's examine the model and categories in Figure 1.
Breaking these components or categories down and examining what's important within each factor is the next step. Again, working under the assumptions of usual importance, a typical model of categories and subcategories is illustrated in Figure 2.
Having run this scenario multiple times, I have generalized a data center site selection study previously conducted based on popular locations for various types of data center users, namely (1) collocation providers, and (2) enterprise users. In the 10-market example as shown in Figure 3, the data entry yielded the results cited.
Noteworthy caveats are as follows: the lower the rating, the more favorable a given market will be for locating a data center based on the attributes and weighted importance outlined in this study; the subcategories act as a catalyst to formulate a categorical value; the quantitative figures cited, for simplicity's sake, have been transformed from numbers to colors, where green indicates favorable, yellow is moderately favorable, and red is less than favorable.
The Ultimate Goal
In summary, each and every user in the market for a data center will have different needs and varying levels of concerns toward those needs. This will modify how one views a favorable location - which will ultimately change the actual favorable locations.
What's important in the process is to bridge the corporate communication gap between those who can provide valued input of both what you currently have, what you'll need in the future, what the important drivers are, and how important these drivers are to you. Oversights, miscommunications, and a lack of internal corporate participation can translate into delays, ultimately costing you money. Therefore, by utilizing the strategy outlined, you can begin to accomplish the ultimate goal of finding the best place to locate your data center.