January 14, 2014

When the financial markets began to sour in 2008 during the peak of the global financial crisis, industry heavyweight Fidelity Investments realized it had a datacenter challenge on its hands.

At the time, Fidelity had 82 datacenter locations in North America, with some of the facilities approaching 30 years old. The sprawling technology infrastructure had grown over many years when revenues were constantly increasing and business units were expanding rapidly.

"We needed to get more competitive on IT costs and services," says Eric Wells, VP, data center services, at Fidelity Investments. "Our costs were growing at an unacceptable rate. We also had to get better at deploying applications. Anytime a business unit wanted to do something quickly, it took months to complete."

With many of the older facilities reaching the end-of-life stage for a datacenter and some newer facilities at capacity, Fidelity needed to find a way to incrementally increase datacenter space without overbuilding for today's needs.

Fidelity brought IT and facilities (real estate) leaders together to work on a solution, with the urging of the company's executives, after they saw a Fidelity data center being built. "We thought there could be a better way to build these facilities," says Joe Higgins, VP of engineering and corporate sustainability officer. The goal was to come up with a data center plan that could cut real estate costs by 50% and IT costs by 20%.

Another goal was to come up with a way to build a datacenter to meet capacity needs, but not overbuild for expected needs five or more years down the road, something most traditional datacenter projects try to do. "It is difficult to predict how much capacity you need two years or four years from now," Higgins says. "It is nuts to build big datacenters the old way. Many organizations are building them, but not filling them."

Fidelity wanted to build a datacenter that could be completed in a short period of time, and then have the capability to expand later as computing needs increased. "One of the core tenets of our project was how to deploy a datacenter quickly," Higgins adds. "We wanted to have responsive capacity and to have it available in months. This allows us to have small bets, so if we miss on small bet, it is not as bad as missing on a huge facility."

Fidelity IT and facilities experts sat down and made "literally thousands of decisions" and eventually developed a datacenter that can be built in stages, off-site, and then moved into place, Wells says.

The resulting product is called Centercore. Centercore allows Fidelity to increase space in 500-kilowatt increments, rather than a traditional large datacenter that would be more than a megawatt. Centercore is being used at Fidelity locations in North Carolina in Nebraska to expand the traditional (raised-floor) datacenter facilities.

Centercore can help a datacenter be built much faster, and cheaper, than a traditional datacenter, Fidelity says. "Traditionally, you would take 18 months to build your own facility," Higgins notes. "We can do it in six months, and 10% to 15% cheaper than regular construction." In addition, if a firm were to build a traditional facility, it would "overbuild" for the firm's current needs, since the costs are so high with a traditional buildout.

"I am completely sold that building in increments is the way to go," Wells says. "The unpredictability that exists with technology and in the financial markets" makes it hard to predict what will be needed in the future. "If you are building a huge datacenter, you are spending a lot of capital. Having the ability to spend in increments and evolve with technology, that makes complete sense."

ABOUT THE AUTHOR
Greg MacSweeney is editorial director of InformationWeek Financial Services, whose brands include Wall Street & Technology, Bank Systems & Technology, Advanced Trading, and Insurance & Technology.