Should You Build a Data Center Today? 2 Universities, 2 Answers

When it comes to building campuses from scratch in the information age, few institutions have a track record like New York University’s. Under its current president, John E. Sexton, NYU has opened campuses in Abu Dhabi and Shanghai. In 2012 it earned New York City’s backing for a new graduate-level institute in downtown Brooklyn.

While in New York recently to cover Cornell Tech and its ambitious plans to build an applied-sciences graduate school on Roosevelt Island, I asked members of New York University’s IT brain trust how they’ve approached similar challenges. Many of their responses mirrored what I heard at Cornell Tech: Don’t focus on individual technologies. Make flexibility a priority. Keep a long-term outlook, even if all the eyes will be on opening day.

Those are principles, though. Some might call them platitudes. What universities actually do often reveals the hard choices embedded in such broader values.

Take data centers, those server-stuffed behemoths that have long been fixtures of campus architecture.

Cornell Tech will not build a data center for the Roosevelt Island campus. That was a major decision, one that seemed to both excite the director of IT, Scott Yoest, and worry him. Cornell Tech figures it can bank on the growing cloud-computing sector to handle its data-storage and research-computing needs. And, by forgoing a data center, Cornell Tech can limit its cost and its energy consumption.

NYU, though, will be building a “multi-petabyte storage facility” alongside its 150,000-square-foot campus in Brooklyn. The university’s IT staff believes the cloud-computing sector won’t expand fast enough to accommodate the voracious data needs of its new institute.

“In our sector there’s not enough market to make it affordable for commercial vendors to provide what we need when we need it,” says Marilyn McMillan, NYU’s vice president for information technology. Adds Thomas Delany, vice president for global technology, “We can’t wait 10 years for the commercial providers to address our vertical. We can’t wait.”

NYU’s new Brooklyn institute, the Center for Urban Science and Progress, is predicated on analyzing big data sets, often highly sensitive ones provided by government agencies. Computing speed and privacy will be paramount. Cornell Tech has more of a start-up bent. It’s less likely that faculty members will be calling on Mr. Yoest to move massive chunks of data or run the kind of burdensome computing exercises typical of a computer-science department. In the event they do, Mr. Yoest says, Cornell Tech can either co-locate with an industry vendor or call on resources at Cornell’s main campus, in Ithaca, N.Y.

Both institutions are looking forward. Cornell wants to capitalize on the growing capacity of the cloud to ease an old burden and increase future flexibility. NYU is measuring that same growth against how its needs will evolve. The disparity in solutions highlights a philosophical debate more and more IT departments in higher education will have to tackle: Is it better to live in the cloud or to keep some of your data nailed to the ground?