<
THE VANTAGE BLOG Ideas, insights, opinions. Get the latest from the world of tech,
data centers, and all things Vantage.

A Prescription for Your Data Center Strategy Part IV

Part IV: The Custom-Built Home – Build Your Own Data Center

 

In the first three parts of our series, we’ve explored the many options for organizations who need to store their data: from public cloud to retail colocation to wholesale. In our final segment, we’ll discuss a fourth option—building your own data center from the ground up.

 

Building a purpose-built data center is a massive jump. This is a good move for large, well-established organizations with growing demand and the need to achieve massive scale. Companies that fit the bill are few and far between today, and have names like Google, Facebook, Apple and Amazon – an elite, upper tier of companies breathing rarified air and facing unique challenges. That being said, plenty of companies continue to build their own reasonably-sized facilities if control and customization are their top priorities.

 

When purpose-built works: The primary advantage to building your own data center is control. Your organization will have complete ownership of the operating environment, including access, power usage, temperature and equipment. The risk of losing your lease is low or nonexistent if you own the building outright.

 

Having complete control over design and management of your data center also means that your organization has the power to keep its facilities cutting edge and to employ the newest techniques in efficiency, cooling, design and power usage. To use Facebook again as an example, the company’s data centers have served as a birthplace and development ground for OCP designs.  Facebook is a shining example of an organization that has built data center design and strategy into a core component of its business and invests heavily in how it builds, operates and manages its data centers.

 

Challenges of purpose-built: But with total ownership comes complete responsibility. There are myriad operational considerations and costs, including uptime maintenance; managing cooling systems; maintaining backup systems and generators; fire suppression and detection; hiring and managing security; building out your own operating procedures and emergency protocols; managing permits, local taxes, power and water usage; and, of course, the equipment itself. Beyond those considerations are the resources it takes to manage each component – security teams, HVAC teams, electricians, mechanics and engineers. The cost of facility and infrastructure maintenance also tends to be unpredictable, which means this route is really only appropriate for organizations that have sufficient cash on hand to tackle unplanned cost overages.

 

When building a private data center, companies identify existing property, purchase or lease it, and customize. Properties run the gamut from empty lots and fields that require building from the ground up or may refer to extant buildings that need to be retrofitted for data center use. Forrester estimates the cost for building out an existing space is about $200 per square foot. For frame of reference, Facebook’s data center in Forest City, North Carolina clocks in at a whopping 300,000 feet. While you’re not likely to need 300,000 feet, at $200 a square foot, you’re looking significant cost.

 

In addition to the hefty price tag, building out your own space takes two to three times as long (an estimated 12-24 months) as moving into a wholesale space (around five months). And owning a private data center doesn’t stop at the build. In fact, construction might be the easiest part. To properly run and maintain a best-in-class data center capable of handling massive data loads with minimal downtime, you’ll need to hire the best IT talent available—something that is in short supply. Trained, qualified and disciplined engineers are a hot commodity worldwide. It’s not just a space you’ll need to build—it’s an all-star team.

 

When to use purpose-built: If your organization is growing into a global power that is stable and profitable enough to afford the huge upfront investment, build-your-own will offer total control of your infrastructure and the lowest operating costs possible. If your company is called Google or Facebook, this makes a lot of sense for you.

 

As more companies expand and rely heavily on Open Compute Project (OCP)—an open-source data center design sharing group made up of industry-leading enterprises, it is becoming increasingly desirable to have complete control by building something to meet your unique needs. However, the expanding markets of managed services, retail colocation, wholesale spaces and yes, even public cloud, are rendering this level of investment unnecessary for the majority. Companies can get 90 percent of the customization and flexibility they desire through a mix-and-match combination of the other services we’ve covered in this series. Even the largest companies still rely on collocated and wholesale spaces, and use public cloud for fast, easy access.

 

Your data center strategy is an ever-evolving undertaking. Understanding the different avenues towards building your customized strategy is essential to landing on the solution that is right for you. Once you grasp the full array of options and the value inherent to each recourse, you will be able to successfully construct a strategy for your company that leverages all of them—from public cloud to retail colocation to wholesale environment to build-your-own. At the end of the day, your data center strategy should resemble the approach you take to your well-being: a well-rounded diet rich in nutrients combined with exercise and self-care is the key to a long and happy life. Draw out your grand design with balance and longevity in mind, and don’t feel that you must choose one option. Rather, mix and match from the choices at hand to create a customized prescription that works for you.

Leave a Reply

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>