Computing coalition may speed big data

For Connecticut researchers dealing with big data, computing is about to get a lot less personal.

Founded in September 2013 by Yale’s Chief Information Officer Len Peters, the Connecticut Coalition for Research Computing (CCRC), a consortium of institutions with an interest in research computing, is in discussions over initiatives to aid researchers who require access to high-performance computing. Since its first meeting last fall, the coalition, comprising members from Yale, the University of Connecticut, Wesleyan University, the Jackson Laboratory and other statewide technology and research leaders, has been investigating potential collaborations that will reduce research costs and encourage resource sharing.

“What’s exciting is that we’re going to achieve collaboratively what it would be much more difficult, even impossible, to achieve individually,” said Michael Mundrane, vice provost and chief information officer at the University of Connecticut. “This is a perfect example of the value of the whole being greater than the sum of its parts.”

In recent years, burgeoning interest in genomics and bioinformatics has generated heightened awareness among institutions of the necessity for high-volume data management. Collaborative high-performance computing initiatives, like those proposed by the CCRC, are already under development across the country. Massachusetts, Ohio, New York, Pennsylvania and other states have also created infrastructures for shared computing, and Peters said the coalition plans to study their models to decide what will work best for Connecticut.

Among the coalition’s considerations is making large data transfers less time-consuming. Peters said the Connecticut Education Network, which provides high-speed Internet access to schools and libraries across the state, has become ill suited for researchers collaborating across universities who need to quickly move big, complex data sets from one institution to the next.

“Right now it’s faster to move data with a truck than with a network,” Peters said.

The coalition has discussed the possibility of creating a data center, modeled on the Massachusetts Green High-Performance Computing Cluster, to reduce the time demanded by computationally intensive, big-data research. Housing the necessary research computing servers in one location, he said, would reduce the distance required for data to travel, and result in faster transfers.

Most institutions of higher learning have their own high-performance computing infrastructures, Mundrane said. But, run individually, these infrastructures cost each institution more than a combined effort would. The coalition hopes to find avenues of collaboration between universities, nonprofit organizations and members of industry that leverage their resources and reduce the financial strain of maintaining expensive technologies.

Beyond serving its researchers, the coalition believes its initiatives align with the economic goals of the state, which include supporting innovative science research. In January 2013, Connecticut governor Dannel P. Malloy announced a $200 million plan to increase Connecticut’s involvement in the biosciences. Given time, the coalition’s work to facilitate research computing has the potential to result in statewide economic development and job growth in the field, and to foster interest from outside researchers said Kiran Keshav, Yale’s director of research service management.

“Any time you can attract top talent, it sets you up for the creation of new and novel discoveries,” Kishav said. “And attracting top talent then begets attracting top talent.”

Other institutions in the CCRC include Yale-New Haven Hospital, University of Connecticut Health Center and the University of Bridgeport.

Comments