My family once had some friends who decided to build their “dream house.” I still remember the day we went over to see it. I was amazed at how much their dreams and mine were so much in alignment. All the kids had their own rooms, there was a massive playroom with its own kitchen, and enough bathrooms that no one would ever have to wait, no matter how long a certain sister took to get ready in the morning. As we drove away I could only look back in envy at the Valhalla they had erected.
Two years later they got divorced. It seems that the home of their dreams was, in reality, a financial nightmare — they went a teeny bit over budget — and the property they built on because of its affordability, resulted in commutes that made the Bataan death march look like a pleasant stroll. I use this cautionary tale as an example of just how wrong things can go in even the most exciting of projects and to help you ensure that your next data center project doesn’t result in acrimony, accusations, recriminations … and with everyone having to wait to use the bathroom.
It Starts With Planning
Abraham Lincoln once said that if he had eight hours to chop down a tree, he’d spend the first six sharpening his ax. What this means to you as the erector of the new home for your company’s data processing and storage requirements is that everything starts with planning. This doesn’t mean that you have to completely design the facility before deciding where to put it, but it does mean you better have developed a clearly understood scope of the project and its requirements before breaking ground anywhere.
Developing the plan for your new facility must take into account a wide array of issues, and the following questions are critical for starting that planning process:
• Why are you building this new data center? As much as it may seem like it, this isn’t a “duh” question. The purpose of building a new facility immediately determines a number of parameters for the data center being constructed. For example, if your site is going to become your primary facility to support your mission critical applications then certain reliability related features are required like a hardened shell and a 2N architecture. Modular data center offerings can’t deliver these types of features, so just by answering the “why” narrows the acceptable options to use in the site’s construction and begins to define its essential requirements.
• What applications will be housed within the facility? Are they critical to the organization and how important is the information they contain in terms of security? Do you need a hardened facility and a 2N architecture or is N+1, or even N sufficient? If you are interested in building a new data center it would seem that the answers to these questions would be yes, yes, and no, but that isn’t always the case. Obviously, no two companies’ data center requirements are alike, but determining what applications the new facility would support enables you to better frame your requirements in areas such as security and reliability.
• What are your projections for the future? In other words, what capabilities must be included in your data center to ensure that it remains a viable asset for the next 20 years? This is an important question since no one builds a data center to discard just five or ten years later, but this happens more than you would think. The nature of the data flowing through a data center is changing. Many applications, video for example, are driving larger, rich packets through facilities while the Internet of Things (IoT) will be generating billions of tiny packets that will need to be processed “instantaneously” if the information is to be used effectively. If you don’t understand your long-range goals you can easily build a facility that will run out of space and/or power long before you wanted it to.
• What are your projected requirements for space and power? This question should be looked at from the perspective of question three above. Things like whether you plan on using high- or low-density racks have a bearing on both your power and space requirements. As you can see, answering questions three and four should be a little more detailed than just white boarding a wish list.
• What is your projected rate of growth? Along with questions three and four, your projected frequency of expansion is a critical element of consideration when you are determining the requirements for your new facility. For example, if you expect to grow quickly you may elect to build a slightly larger data center than dictated by your initial requirements, or since most data centers expand in larger “chunks” do you need a facility that is designed for easy expansion without interrupting your ongoing operations?
This is certainly not a comprehensive list of the questions that need to be addressed in the planning and design phases of your project, but as you have probably already deduced, your answers will generate new sets of issues that will need to be addressed. In other words, questions will beget questions and the answers will demonstrably affect the site selection phase of your project.
Selecting A Site
Unless you are fortunate to already own the land where you want to locate your new data center — and as you’ll see, even that is not a guarantee that it will support the facility — site selection is a little bit more complex than simple saying “we want to put it here.” In selecting your site, here a few of the key guidelines to keep in mind:
• Site selection is a process of attrition. A data center site isn’t so much selected as much as it is often the last one standing in a process of elimination. When factors like access to utilities, fiber, existing sites (synchronous communications encompasses a range of only 30 miles), local regulations, and proximity to performance affecting entities like a chemical factory are considered the “perfect” site often isn’t quite so perfect. This process of elimination also necessitates that you understand the zoning in the areas surrounding your dream location to understand any special requirements that you may have to take into account. We once built a facility near a residential area where we needed to dynamite through some rock formations. We were able to do it but we needed to put up the proper precautionary signs and speak to the surrounding neighbors to inform them of the times that these actions would be occurring.
• You have to understand the local “layout.” No two municipalities are the same. What is acceptable for one may not be in another. This may even be found to be the case in adjacent municipalities. For example, in building two new data centers in adjoining cities we had to build extensive baffling to reduce generator noise in one city while its neighbor had no such requirement.
• Obtaining permits and approvals is an arduous process. This issue is an extension of understanding the local layout. Even if you’ve purchased the land that doesn’t mean that you can build a data center there. The permitting and approval process can take months and even then, there is no guarantee that your project will be approved. Issues surrounding permits and approvals are also a major reason why using an existing building isn’t necessarily faster than building from the ground up.
Understanding the local “layout” is also important here since familiarity with the unique elements in a location’s processes can dramatically reduce the time required to complete this phase of your project. While the basic requirements tend to be the same, different communities often place more emphasis on specific items. One city that we built a data center in was very concerned with the aesthetics of the building itself and we had to provide very detailed renderings of what the completed facility would look like. When we were asked to build another site in the same community, we shaved more than a month off of the permitting process by preparing the required renderings in advance. “Be prepared” isn’t just a motto for the Boy Scouts.
• Beware of hidden requirements. Just because you want to “just build a data center” doesn’t mean that your selected site won’t have to include requirements (like what should be included in your landscaping) that you haven’t anticipated or, more importantly, budgeted for. In one municipality we were required to build a “community gathering” area as part of the project. Just when you think you’ve seen everything, you can quickly learn that you haven’t. The gathering area turned out to be quite nice however.
• Experience counts. Site development is a learned skill. On the job training is not the methodology you want to use when selecting and developing a data center site. Attempting to perform the activities necessary to perform this function with personnel, consultants, or contractors who are not familiar with data centers will only add time and costs to your efforts. For example, you’ve acquired the land only to find out that a gas main easement runs through it or the site has actually been built but someone neglected to send in the load letter and the utility dedicated the power elsewhere. If you don’t have this experience within your organization you will need to retain an expert in this area to ensure that your development schedule doesn’t become measurable in years.
Although building a new data center and finding and selecting the site where it will reside can be a detailed process, the most important element in successfully navigating all of the associated elements is to do your planning first. There is simply no substitute for a detailed and well understood set of requirements to alleviate the various pitfalls associated with any data center project. Knowing what you want, and what is needed to get and support it, seems like overly simple advice, but a majority of data center projects are initiated with little more than the equivalent of a “wish list,” and oftentimes those dreams wind up being more like nightmares.