Scaling Up Vs. Scaling Out In Data Center Site Selection
A new approach to data center design.
Today’s emerging technologies — big data and analytics, cloud computing, Internet of Things, and others — are pushing the limits of density, power, and space in the data center and forcing data center providers to re-think every facet of data center design.
In fact, there’s a growing trend of data centers using design to address opportunities in capacity management, modular deployment at scale, challenging environments, energy efficiency, extreme density, and target opportunities in edge markets and cloud providers.
Many design innovations — such as new temperature and humidity guidelines, renewable energies, liquid cooling, the ability to better monitor and manage environmental conditions via data center infrastructure management (DCIM) software, and open standards — are driving down costs and ushering in needed efficiencies.
The truth is that data centers and their designs are constantly evolving to meet the fast-changing needs of customers and their applications and data. But the most profound impact of data center design starts with the site selection process.
Creativity In Site Selection
Today’s data center providers are pushing the boundaries of what a data center can and should be — and they’re using the site selection process to do it.
Foxconn’s latest green-tunnel data center is built inside a long tunnel within the Guiyang industrial park in China to take advantage of wind energy. There are also data centers located inside a former nuclear bunker, a 1920s Spanish chapel, and a range of former industrial sites that have been converted to modern data centers. As Arthur Cole has previously commented, “Now that physical constraints have been lessened, the only thing stopping innovation these days is the human imagination.”
Key Considerations In Site Selection
But it all starts with site selection. A well-defined site selection process considers a number of factors: the environment, climate, power, connectivity, taxes, land, economic incentives, proximity to transportation hubs, construction and permitting, labor, and other factors.
For example, when Intel needed to expand its data center operations to support its design of silicon products, they turned to an established set of site-selection criteria that they use to maximize their return on investment. These Intel criteria include:
- Environmental conditions, including climate and history of natural hazards.
- The availability and cost of fiber and connectivity.
- The availability and cost of a power infrastructure.
- Site-level criteria, such as land acquisition, proximity to threats and resources, and the construction environment.
- Socioeconomic, workforce, and governmental criteria (such as regulations, taxation, and incentives).
These criteria help them choose a site that optimizes construction and costs, meets internal customer requirements, and accommodates future expansion. This well-defined site-selection process results in a choice of sites that offers the most advantages throughout the life of the data center.
Scaling Out Vs. Scaling Up
While data center site selection often focuses on the location of the data center itself (in a wind tunnel, chapel, etc.), it can also result in new ways of thinking about the data center footprint.
For example, data centers have traditionally scaled out to accommodate its expansion needs. While scaling out works well in rural areas, where land is plentiful and cheap, it can be much costlier for urban data centers where property and taxes tend to be much higher.
We faced this very problem at our IO.Phoenix data center campus. Our customers’ growing needs for modular colocation and cloud services required us to think bigger. So we were able to secure a nine-acre parcel behind our current 530,000-sq-ft Phoenix data center on which to build a multi-story data center facility.
The three-story building will feature approximately 200,000 sq ft on each floor for a total of 600,000 sq ft of space. As it is filled, the new data center will grow to 100 megavolts of capacity. Upon completion, we will have more than 1.2 million sq ft of data center space on the Phoenix campus.
This multi-story facility will allow us to house more colocation modules within a much smaller physical footprint.
The Logistics Of A Multi-Story Data Center
In addition to housing more modules in a smaller footprint, the new data center will place its equipment yard on the rooftop of the building to maximize the use of space. IO has leveraged this expertise outside the U.S. and will incorporate it into its multi-story data center design.
As new modules are manufactured and shipped to the data center campus site, they will be raised and positioned inside the building. The 42-ft long modules are then moved using “air skates” that use compressed air to raise heavy loads and will allow one or two employees to move modules as heavy as 20 tons.
Clearly, innovation in the data center is being pushed to new heights. Those that capitalize on innovation will be well positioned to meet the growing needs of customers and their data.