The reality is that every company won’t need what’s known as a Tier IV data center—the most robust and redundant of all the tier categories. Depending on their needs many companies can get away with Tier I, II, or III centers, which have varying levels of functionality and redundant components. Organizations that typically need Tier IV centers are in banking or health care, where even a few minutes of center downtime can do significant damage, or the need for data capacity is higher to comply with regulations such as Sarbanes Oxley.

To help control costs, Hansen will sometimes recommend that clients make part of a data center Tier IV—perhaps 25 percent of space—and build the rest of the space to a lesser standard. “We try to break a client’s environment down into modules, and decide which need to be at higher tiers and which can be lower,” he says. “The cost difference between Tier I and Tier IV is considerable.”


Bridging Cultures

Hansen’s experience with these complex facilities proved handy when he worked on construction projects in Bangalore, India, a place where it’s common for power to get knocked out eight to 10 times per day (in contrast, U.S. data centers typically lose power about two to three times per year).

“The emergency back-up infrastructure there isn’t used for emergencies but rather is put into everyday use, so the type of generators and UPS (uninterruptible power supply systems, which are designed to eliminate short-term power outages) in use are far more vital, and you have to take a different approach in how you design them,” Hansen says.

Information security regulations also differ within India, which influences, for example, decisions made by U.S. companies with a presence in the country. “We have laws in the U.S. that protect confidential client information that’s on back up tapes, but it’s not true everywhere in India,” Hansen says. Such differences mean companies have to decide whether to decentralize servers and place them overseas, or use a centralized model where employees can connect to servers based in the United States.

Designing data centers also means balancing the competing interests of two groups. The facilities management group usually owns the physical building and is responsible for paying power and cooling bills. But the information technology department, which oversees technology inside the structure, has the biggest influence on the size of that power bill through its choice, management, and use of hardware. That can be a recipe for conflict.

“Often my biggest challenge is getting those two groups to be on the same page,” Hansen says.

Some companies have chosen to merge or reconfigure responsibilities to try to skirt turf battles, with the result sometimes being that IT becomes accountable for power and cooling costs. “We have seen some good success around that arrangement,” Hansen says. “When you can get those two silos talking and cooperating, good things can happen in data center design and management.”