Launch of Shared Research Computing Facility


December 17, 2013

On October 29, Columbia University held a ribbon-cutting ceremony for the launch of the new Shared Research Computing Facility (SRCF) and a 1600-core high performance computing (HPC) cluster serving the Morningside and Lamont research communities. This new facility will consolidate computational resources and improve data storage options for diverse research groups across the University.

The NIH, the Empire State Development’s Division of Science, Technology and Innovation (NYSTAR) and the New York State Energy Research and Development Authority (NYSERDA) were acknowledged at the ceremony for supporting these important additions to the University.

"The launch of the SRCF and the new shared HPC cluster in October is an encouraging step towards meeting the growing computing needs of Columbia researchers," said Rajendra Bose, Manager of Columbia University Information Technology (CUIT) Research Computing Services.

“Researchers are able to better focus on the goals and challenges of their research, while the Data Center staff ensures that the technology operates to specifications,” Candy Fleming, CUIT Chief Information Officer and Assistant Vice President said.

The new HPC cluster is a joint purchase and partnership between 10 research groups and departments, CUIT and the Office of the Executive Vice President for Research. The cluster is also supported in part by Arts & Sciences, The Fu Foundation School of Engineering and Applied Sciences and New York State. The SRCF will house the HPC cluster and other, future research computing infrastructure that will enable faculty, staff and students to further advance the University’s research mission.

The SRCF, which is a dedicated portion of the University’s Data Center, is an efficient central location for shared clusters because of the power and cooling requirements of the hardware systems used in high performance computing today. With the evolution of computational research methods and the growing use of modeling, simulation and analyses across many fields, universities like Columbia are assuming a larger role in supporting centrally-managed research computing resources.

CUIT manages the technical infrastructure of the SRCF on behalf of the University’s researchers. The specification, purchase, installation, configuration and ongoing management of the facility has been the responsibility of a number of CUIT groups, including systems, storage, network engineering and operations, data center facilities, research computing services and desktop and software engineering.

“The collaboration within the CUIT organization was vital to the success of this project and required the involvement of several infrastructure teams,” said CUIT Director of Research Services, Halayn Hescock.

In 2008, with support from the Office of the Executive Vice President for Research (EVPR), researchers from the Departments of Statistics and Astronomy & Astrophysics agreed to pilot a shared HPC cluster (“Hotfoot”) to be purchased, installed and managed by CUIT. Prior to this, most departments purchased and ran their own computer systems for research purposes.

“The partnerships that we built with CUIT, EVPR and the participating research groups will continue to strengthen and build as we look to the future in high performance computing,” Halayn Hescock continued.

While the use of computational resources among Columbia’s researchers varies significantly, there are few researchers who use 100 percent of their clusters at all times. The sharing of the HPC cluster offered a more appealing alternative: Access to potentially greater computing capacity, when available, while eliminating the burden on researchers of managing their own systems. To allow the cluster to be purchased, installed, housed and administered by CUIT staff was clearly the best use of resources. It also helped further contribute to the University’s goal of reducing its carbon footprint. Within two years, the cluster doubled in size.

In 2010, the University received a National Institute of Health (NIH) grant that was used to upgrade the electrical infrastructure of the University’s Data Center so it could house the SRCF. “The funding from the NIH grant enabled upgrade of essential electrical infrastructure required for all Data Center operations, supporting advanced research as well as education and administration across the University,” said Candy Fleming.

The University also received a grant from NYSTAR to contribute to the capacity of the 1600-core HPC cluster. With the shared goal of advancing research at Columbia and the funding in place, the project was realized and launched in October 2013.