The Role of Grid Builders in the World

ll of us who work at the task of building the computational infrastructure we call the Grid must regularlypause and remind ourselves of the real nature of the responsibilities we carry. What is at stake in building
the Grid is scientific and technological progress itself: the prospect of producing enough food for 6 billion people and the hope of understanding and curing a host of diseases, to say nothing of resolving the paradox of finite environmental resources confronting an apparently infinite need for energy. While it is not the answer to everything, the Grid certainly represents the promise of finding solutions for a needy world.

Thinking, either globally or locally, involves the collection and processing of information, the very processes the Grid is enabling at the highest level. It involves linking researchers, their instruments, their data, and their visualizing capabilities, seamlessly. Fortunately, the Grid builders are legion and there appear to be no insuperable obstacles. All the major institutions of our society--universities, industries, and government--are participating. The two NSF Partnerships for Advanced Computational Infrastructure are well aware that the enterprise of Grid-building is large and loose, has no defined boundaries, and depends heavily on the initiative of individuals and groups who find a need and fill it. Our view of the way in which the Grid builders should proceed can truly be expressed in one word: collaboration.

GRID PARTNERS

THE BEST POSSIBLE INFRASTRUCTURE

As "start-ups" three years ago, NPACI and NCSA collaborated and learned to do exactly those tasks that should be transparent under an advanced computational infrastructure.

GRID PARTNERS

Collaboration was the whole idea in starting the NSF PACI partnerships, of course, and in our experience it has been central ever since. As "start-ups" three years ago, NPACI and NCSA collaborated closely in moving data archives and assisting supercomputer researchers to begin their work anew on unfamiliar machinery. In the process, we learned to do exactly those tasks that should become transparent under an advanced computational infrastructure.

Today, we collaborate not only across our partnerships, but also with investigators and programs of many national agencies, taking roles in the NSF Digital Library Initiative, elements of the Department of Energy program, the NASA Information Power Grid, National Institutes of Health, and such organizations as the Federal Web Consortium and the Grid Forum. We have implemented a combined allocation system through our National Resource Allocation Committee to make all our resources easily accessible. Our joint education and outreach objectives are pursued through a common organization, the Education, Outreach, and Training Partnership for Advanced Computational Infrastructure (EOT-PACI). We work together to foster computer science developments particularly emphasizing HPC. Prime examples are the development and deployment of metasystems such as Globus, data mining technology, digital libraries, and programming environments for parallel computers.

THE BEST POSSIBLE INFRASTRUCTURE

Another example of NCSA and NPACI collaboration is the National Laboratory for Applied Network Research (NLANR), funded by NSF's Advanced Networking Infrastructure and Research Division. NLANR's principal investigators, from NPACI, NCSA, and Carnegie Mellon University, have been close collaborators for a dozen years. They have played a major role in defining the basic measures of network performance, and in the most applied manner, they have delivered invaluable measurements, analyses, and engineering assistance to the builders of high-performance networks. Our initiatives in data-intensive computing and other enabling technologies have all benefited from this central wisdom of collaboration.

NPACI and NCSA collaborate because the mission is to achieve the best possible national computational infrastructure--not just to develop two pretty good infrastructures. Do we also compete? In a sense, we do. An example is evaluation of diverse high-performance computing architectures.

While our work is only a small part of the whole effort, the leadership of this work is not a small responsibility. Collaboration will continue to drive our approach to new national initiatives in information technology and to other opportunities that will arise. We compete when the best interests of our communities are well served by competition. And we collaborate to make sure those interests are well served within the ever-changing structure of American science and engineering. A ubiquitous, continuous, pervasive, and dependable computational infrastructure will, like a rising tide, raise all the boats of science and engineering so they may see and steam toward the farthest horizons. *