Chapter 4: Thinking
66 - The Grid (Part One)
The first big test of what is now called “grid computing” debuted at a San Diego supercomputing conference in 1995. There, eleven high-speed networks were lashed together for three days to create a single meta-computer. Through it conferees could, among the many cool tools on display, try out a working ecosystem model of the Chesapeake Bay and watch spiral galaxies collide. As one of them recalls, “It was the Woodstock of the grid—everyone not sleeping for three days, running around and engaged in a kind of scientific performance art.” The project was called I-Way, and it launched what many feel is a major step forward in the evolution of the Internet. In that view, the first step came when computers were hooked together by wires to produce the Internet. The second step was the World Wide Web, which created a common interface, a way of sharing information over the Internet. Now with the third big step—Internet grids—we can share processing power itself.
Just as with the World Wide Web, grids rely on a common interface—in this case the Globus software designed by Ian Foster, a senior scientist at Argonne National Laboratory, and Carl Kesselman, who heads the Center for Grid Technologies at the University of Southern California (this work was funded by DARPA). Their decision to make Globus open-source—to publish its code for free on the Web—has spurred the rapid rise of grids around the world.
And that rise is truly rapid. Globus was released in 1996. Just three years later in the U.S., the National Science Foundation and the Department of Energy set up their Access Grid—to serve large, multiple-site meetings online—while NASA deployed its Information Power Grid for aerospace and planetary science. Ensuing years brought NSF’s Grid Physics Network, Germany’s Unicore, the U.K. National Grid, the European DataGrid, and the International Virtual DataGrid Laboratory—the first true global grid, linking supercomputing centers in Europe, the U.S., Japan, and Australia into one behemoth processing entity.
The power of grids is growing, too. In the United States the new, $53 million TeraGrid links four supercomputing clusters into a transcontinental PDP that can process twenty trillion mathematical operations per second. A project called Enabling Grids for E-science in Europe will build the largest international grid so far, linking seventy institutions into a twenty-four-hour grid service with power that equals twenty thousand large computers combined. And the new power is being used: for efforts in climate modeling, gene research, high-end physics, brain studies, even earthquake modeling. It’s also being taken up by business.
Email this to a friend | 
Add to del.icio.us | 
Add to digg |
Linkology by Kate (all links are unofficial) | Rights Info
