Globus Alliance bags $13m NSF grant


Author: Tina Gasperson

The National Science Foundation, which is conducting several long-term experiments that rely heavily on open source grid computing tools produced by the Globus Alliance, this week pledged more than $13 million in long-term support to the Globus development community.

The Globus Alliance is a group of computer scientists, led by Carl Kesselman of the University of Southern California Information Sciences Institute and Ian Foster of the Argonne National Laboratory, University of Chicago, working to make grid computing a “realistic possibility in science,” according to the Alliance Web site. The Globus Alliance hopes to establish a standard for grid computing that will bring forth what it calls “the Grid,” marked by a time when every system on the network uses the same protocols, much like the hypertext-based collection of nodes on the Internet is today called “the Web.”

The Alliance’s flagship project, launched in 1996 and dubbed the Globus Toolkit, is very popular in the scientific community and has been called the “de facto standard” for building grids by the New York Times. DARPA, the Swedish Royal Institute of Technology, the UK e-Science Grid Core Programme, the National Science Foundation (NSF), the US Department of Energy, and several other entities have given grants to fund continued development of the Globus Toolkit. “We have received a little bit of money from the EU and some commercial money,” says Kesselman. But recently, the Globus Alliance decided it needed a more long-term funding solution. Members put together a grant proposal and approached the NSF, which has been making heavy use of the Globus Toolkit in several projects, including the GRIDS Center Software Suite, part of the Foundation’s huge Middleware Initiative (NMI), launched in 2001.

“Some of our other funding streams had gone away because of programs ending and refocusing of activities,” says Kesselman. “And we previously have been supporting this thing with five or six or seven different overlapping grants.”

The Globus community wanted to eliminate the need to fulfill reporting requirements for so many sponsors, while the NSF wanted to ensure continued support and development on a project it had come to depend on. So the agency agreed to provide a generous $13.3 million grant over five years. “This kind of rolls it all up into a single source,” Kesselman says. “Some of these scientific projects have far-reaching horizons — some of the large experiments might have life spans of 10 years, so its important for them to have assurances that there will be support and bugs will get fixed and things will advance.”

The Globus Toolkit would have survived without the money, Kesselman says, but at a greatly diminished level. “We would have figured out ways to keep going but it would have been incredibly difficult. [NSF’s] alternative would have been to do it by themselves, which would have been much more expensive. It’s a win for everyone. We get to continue advancing the technology and the scientific community can focus on what they want to do.”