Blue Gene Q Gives the University Computing Muscle
Deep within an unassuming-looking building just off campus lies the Research Data Center—the brain that supports Big Data at the University of Rochester.
The center's computing capabilities are mind boggling. The systems housed here have an aggregate computational capacity of 240 teraflops or 240 trillion calculations a second. That's the equivalent of more than 20,000 laptop computers.
The research facilitated by all this computing muscle ranges from the study of young stars in distant galaxies to simulations of the human heart.
But Big Data wouldn't be such a big deal at the University if somebody didn't maintain those big machines and—even more important—provide the training and support to the researchers who need access to them.
That's where the Center for Integrated Research Computing (CIRC) comes in.
CIRC supports 550 users with individual and group training, and it hosts monthly symposiums where researchers can showcase their work, learn about emerging computing technologies, and participate in collaborative discussions.
Those users, by the way, include faculty members, postdoctoral scholars, research staff, graduate students, and undergraduates from more than 35 River Campus and Medical Center departments and centers. "We support the computational needs of researchers across the University community, including high-performance computing and big data applications," says director Brendan Mort, who heads a staff of six.
Most of those staff members, he adds, are located in Taylor Hall. "We are located with researchers on campus—rather than with the machines at the RDC—to emphasize the importance of collaborating with faculty, students, and research staff and providing them the assistance they need in using the systems," Mort says.
The Research Data Center began to coalesce in 2005 out of recognition of a need for shared computing resources. A faculty team of 17 researchers recommended not only the purchase of a large high-computing cluster but also formation of a center to provide support and training. What was then called the Center for Research Computing received joint funding from the College and the Medical Center in 2007 and was officially launched the following year after purchase of the Linux cluster.
Later that year, IBM donated a Blue Gene/P supercomputer as part of its partnership with the University in establishing the Health Sciences Center for Computational Innovation. The success of that center led, in turn, to purchase of the Blue Gene/Q in 2012 with state funding awarded at the recommendation of the Finger Lakes Regional Economic Development Council.
Now that all that computational power is in place, the next priority is people power. "We need more translators—computational scientists—to work with the scientists in the labs to do the high-performance computing that's needed to analyze huge amounts of data," Mort says.
CIRC maintains three major computing systems—the flagship, a Blue Gene/Q supercomputer that accounts for 209 of those teraflops of computing muscle, an NX cluster that allows researchers to use high-performance computing and Big Data applications remotely in their labs with the ease and convenience of a desktop environment, and a Beowulf-style Linux cluster called “Bluehive”—in acknowledgement of the school’s mascot, the yellow and blue Yellowjacket wasp called Rocky.