The Grid Research to Enterprise Funnel

GCJ: Could you tell us something about the Computation Institute, and your new role there?

Foster: The Computation Institute was established as a joint venture between the University of Chicago and the Department of Energy's Argonne National Laboratory-which the University has operated, by the way, for 50 years, and will operate for another 20 years at least under a new contract recently awarded. Its mission is to "address the most challenging problems arising in the use of strategic computation and communications."

Under the able leadership of Rick Stevens, the Institute has grown substantially since it was founded in 2000, hosting major initiatives such as the Grid Physics Network project (GriPhyN), TeraGrid, the National Microbial Pathogen Data Resource, and the Social Informatics Data grid (SIDgrid).

I assumed the role of director in March of this year. I agreed to take the job because I believe that for the Computation Institute, the past is but prolog. We have achieved fine things, certainly, but we also have the opportunity to do far more. I see us as being on a cusp in terms of our ability to exploit new research approaches based on large-scale computation, data, and communications. The combination of a world-class university, oriented towards fundamental intellectual pursuits (Chicago has more Nobel prizes than any other institution), and a world-class federal research lab, focused on mission-oriented science and engineering, creates many interesting opportunities to pursue transformative computation across a range of disciplines. And I've found that the leadership at Chicago and Argonne is well aware of the importance of developing the capabilities required to exploit those opportunities.

GCJ: So the Computation Institute isn't just about grid?

Foster: Indeed not. We're concerned with understanding and applying large-scale-or, as we like to say, strategic-computation, data, and communications wherever they can be used to advance knowledge. This mandate certainly encompasses grid, but also other technologies, including high performance computing. I should also add that we're concerned not only with the physical and biological sciences, but also social sciences, engineering, medicine, and the arts and humanities-all areas of strength at Argonne and Chicago.

That said, I do see grid as underpinning much of what we do. Whether scientists are producing large datasets or new simulation tools, running detailed numerical simulations, performing large-scale data analyses, or creating new scientific instruments, they inevitably end up requiring community input and/or producing assets that need to be made available to the community. Thus, essentially every problem that we address will involve large-scale collaboration, and thus communication and grid technologies.

GCJ: And how many people do you have on board at this point?

Foster: We have over 70 faculty and scientists at Chicago and Argonne who are Fellows of the Computation Institute. And over 80 people - students, staff, faculty - are paid in whole or part by our various research contracts.

GCJ: Is everybody using Globus at some level? Or are there a number of different tools you and your researchers use?

Foster: Essentially everyone who's involved with grid at Chicago and Argonne is working with Globus. All of the major grid projects that we're involved in, TeraGrid, for instance, as well as the Open Science Grid, are Globus-based. Of course, some other projects are concerned with parallel computation, or machine learning, or with theoretical aspects of computing, and in those cases grid technologies are not employed-at least not yet!

GCJ: I see. And who funds the Institute?

Foster: At present, our projects are funded primarily by the federal government. However, I believe that this situation needs to change to a significant extent if we are to realize the full promise of computation for science. Thus, to jumpstart the Computation Institute's expansion, I've obtained substantial new internal funding to support the hiring of new faculty and staff. And we're starting an aggressive fundraising campaign to endow the Institute with the ability to take on problems on the basis of intellectual merit, rather than solely the availability of external funding.

We're also talking increasingly with people in industry with whom we have a common interest. That has not been a focus to date, but I think will be important in the future.

GCJ: Interesting. Any chance that tools developed within the Institute will be released to the public?

Foster: Yes. One thing that I will certainly be emphasizing is spreading the tools we build to the community, rather than sequestering them for our own use, because I think that's ultimately the way that our faculty and staff will have the greatest impact on their peers.

It's worth mentioning in this regard that the process by which tools are made available is changing rapidly, from code you download to software you access over the network: from tarballs to services if you like. I hope that we will be leading in what really is a revolutionary change in the practice of computing and science. There are big challenges to be addressed in that regard: it is certainly not easy to take something like a genomic database and its related analysis tools, and put them up as a service that can be accessed by thousands, or perhaps tens of thousands, of users.

GCJ: Right. There is an enduring argument in some portions of the grid community that grid is very much for research, and academia-the point being it hasn't made the leap to the enterprise. Do you think organizations like the Computation Institute are likely to play a role promoting grid to the business community?

Foster: Absolutely, yes. And we've got the opportunity to make contributions in a few different ways. One is by building tools and developing techniques that are directly useful. Another is by demonstrating feasibility. I think there's also the opportunity at a place like Chicago to contribute to thinking about how this technology should be used in industry. We have a first-rate Economics department and business school, and we are getting some of those people involved in thinking about what this technology means for business and society in general.

GCJ: Do you see parallels between the ways in which you and your researchers approach problems and the methods of the business community? I'm trying to understand the mindset on how you guys work, and how that might compare with the activities of the private sector.

Foster: I think the concerns are similar in important respects. In both cases, you have groups of smart people trying to solve difficult problems. Now the problems being addressed at Chicago or Argonne will tend to be different than in a typical company. For example, we might want integrate a large amount of genomic and proteomic information to infer new information about the genetic causes of disease, or information from many literary texts to infer new information about patterns of authorship. In a company, on the other hand, the goal may be to integrate financial information from many sources to manage risk on the stockmarket, or to perform and compare a large number of simulations when designing a new chip. But the process and concerns are not that different. There is a need to maximize the creativity of individual knowledge workers, to allow teams to form and work together effectively, to provide the data and computing resources that workers need when they need them. In fact, I would say there are more and more similarities between how science is done and how business operates, in this regard.

GCJ: Any new announcements or new projects that you'd like to mention before we go?

Foster: We're hiring, so contact me if you if you share our passion for computation and science. For other developments, stay tuned. We're just getting started!

close window