NSF cyberinfrastructure grant will provide dedicated lane for research traffic

The 5-mile drive from Rice University to Houston’s Galleria could take as long as 40 minutes if you stick to streets like University Boulevard and Westheimer. But take the freeway, and you could be shopping in as little as 15 minutes.

A new grant from the National Science Foundation (NSF) Division of Advanced Cyberinfrastructure will give Rice researchers a computer network analogous to the freeway system.

The two-year, $500,000 grant will be used to upgrade equipment, which will allow specific research traffic unfettered access to big data research sets and instruments around the world and expand and improve resources for faculty and students performing scientific research.

Klara Jelinkova

“University networks are very complex,” said Klara Jelinkova, vice president for IT, chief information officer and a principal investigator (PI) on the project. “They serve research, education and administrative functions. These areas have different requirements and increasingly are mandating sophisticated security approaches in order to secure information. These required security improvements introduce latency in the network flow that can negatively impact research. The grant enables Rice to purchase equipment that allows specific research traffic to bypass certain security controls to improve research collaboration. Many of our peers have such equipment, and having this capability is a competitive advantage for research grants in certain fields.”

According to the proposal abstract, the project will support 100-gigabytes-per- second flows between data transfer facilities at Rice’s off-campus data center and national and international research and education data repositories. Rice’s current flow rate is 10 gigabytes per second.

Jan Odegard, a co-PI on the project, gave an example of how this could impact Rice: “Today, our network could, if it delivered peak performance, permit 10 people to simultaneously stream an uncompressed high-definition movie (movie-theater quality) live. After we upgrade, 100 people could do the same.”

Jan Odegard

It’s actually more profound than that, said Odegard, who is associate vice president for research computing and executive director of the Ken Kennedy Institute for Information Technology. Currently, the network is supporting not only research traffic but also administrative and student users. “Today’s network would likely only support three to four concurrent streaming movies, while the new network, since it is dedicated and optimized, could probably sustain more like 90 concurrent streaming movies,” he said.

The upgrades will be important for researchers like Rice physicist Paul Padley who work with large data sets of flows of data from instruments such as the Large Hadron Collider.

“In particle physics, we typically need to move about 100 terabytes of data from national labs to campus to perform a typical data analysis,” said Padley, professor of physics and astronomy and a co-PI on the grant. “Once that analysis is done, we discard that data (there is a copy at the national lab) and proceed to work with a different 100-terabyte data set that we again move onto campus. By having this data fast lane, we can move data on campus more quickly and thus will be much more productive in our output of important physics results.”

Other co-PIs on the project are Keith Cooper, the John and Ann Doerr Chair in Computational Engineering, professor of computer science and of electrical and computer engineering, and Moshe Vardi, the Karen Ostrum George Distinguished Service Professor of Computational Engineering, professor of computer science and director of the Ken Kennedy Institute.

“With faculty increasingly involved in data-intensive research, it is critically important that we develop our network connectivity and bandwidth to support that,” said Odegard, who will be the overall project lead of this collaborative effort.

With this grant, he said, “we will improve by an order of magnitude in raw bandwidth, but we expect more than an order of magnitude can be realized because this is a dedicated lane on the ‘autobahn’ that is optimized for research flows.” As a result, he said, “we will be able to optimize the routes to use a larger percentage of the peak.”

Equipment upgrades are expected to begin this summer.