When you plug into an electrical outlet, you may be using electricity generated in the West and shipped through the South before it reaches your office in the Northeast. Such seamless delivery is made possible by a grid system that transports power where it is needed, regardless of where it is generated or how it is being used.
Now imagine turning on your office computer and tapping into the computing power of a PC in the office of a colleague at Berkley as well as hundreds of other computers across the country with the same ease.
Welcome to the world of the computing grid, which researchers say is the next evolution in the power of the Internet.
Binghamton University computer science Assistant Professor Michael Lewis is among those who are developing this new distributed computing system that promises to bring together scattered computing resources, integrate them, manage them, and take advantage of them as if they were one huge, virtual computer.
Lewis’ efforts were recently recognized by the National Science Foundation, which has awarded him a $395,000 grant from its prestigious Faculty Early Career Development (CAREER) Program for young researchers. Over the next five years, the funds will support Lewis and his research group in their efforts to make changes to grid systems dynamic and automatic.
“Grid computing is not a reality yet, but it’s something that lots of people are shooting for,” said Lewis. “This is a very big initiative for open scientific research.”
Grid computing has attracted the interest of such computing giants as IBM, Microsoft and Oracle and is the focus of intense research interest at NSF centers and universities because of its huge potential to advance computing capabilities.
In effect, the Grid will make the Internet an automated computing platform, Lewis said. This “virtual computer” will provide disk storage for users, transparently swap files and optimize users collective performance, all with no central administration. This means that users will have access to a host of resources through their personal computer.
“The technology is important for research because with that kind of computing power, we could solve bigger problems faster.” Lewis said.
Lewis said grid technology will enable scientists from different parts of the world to collaborate on complex projects that require lots of computing firepower including things like climate modeling, high-energy physics, genetic research and earthquake simulations.
For example, one of the main projects being pursued at the University of San Diego is brain mapping research, which is critical to better understanding and treating Alzheimer’s, schizophrenia, Parkinson’s and a whole set of neurological diseases. The problem now is that each MRI or map of a brain uses huge amounts of storage space. And because many images of different brains are needed for comparison, research must rely on collaboration among many centers, so each can share their data. Grid computing systems will make that possible.
According to Lewis, breakthrough discoveries depend critically on computational and data management infrastructure as a scientific tool.
Grid computing systems promise to be far more powerful and flexible than any single super-computing system. Because of Lewis’ research, BU has become a member of an experimental grid environment, called NPACI-Net, using 200 or so machines connected over a high-speed network. Such experimental labs are the newest trend in supercom-puting.
NPACI-Net runs a grid software called Legion. “With a single Legion account on NPACI-net, researchers here at BU can now build and run their scientific applications on this grid,” Lewis said. “Their programs may be executed in San Diego, Virginia, Minnesota or any of the other machines that make up the grid. The grid itself decides where to execute the programs and automatically returns results back to applications programmers.”
One of the challenges in developing these super systems, however, is to make them operate automatically, making adjustments and updates as needed. Lewis’ research is involved with the process of changing running programs. “What I’m working on is allowing grid objects to evolve their behavior on the fly, so they are able to change or update without having to shut down or restart,” Lewis said.
According to Lewis, this is important because for some applications it is inconvenient or impossible to shut down. “Some components of the grid need to be available 24-7,“ he said. “If you have to shut an application down, it would cause a disruption in service and inconvenience users. What I’m working on would allow changes and updates to be performed seamlessly.”
Much like the change from telephone operators to automatic switching systems for the telephone industry, grid technology will require less human intervention and be able to take care of its own needs.
“If it needs more computing power, it goes and gets it, and doesn’t keep looking for a human to help,” Lewis said.
Lewis’ award is the first NSF Career award received by the BU Computer Science Department. Career awards recognize and support the early career-development activities of those teacher-scholars who are most likely to become the academic leaders of the 21st century. Career awardees are selected on the basis of creative, career-development plans that effectively integrate research and education within the context of the mission of their institution.
“The award will allow me to hire and attract talented graduate students to work with me on this project,” Lewis said. “And attracting the best students is an important component to the success of this project.”
Lewis was also one of the original researchers to submit meritorious projects for the BU’s Internet2 grant proposal in 2000.