"The Grid," as it is often called, has bypassed buzz and entered the mainstream. Named for the delivery model used by the electricity industry, grid software is basically a middleware layer that connects users with unused processing power from a distributed and interconnected network of internal and external computing resources.

"The idea [behind grid computing] is that an intelligent piece of software will instantly know what computer is available at any point in time," said Charles King, a Sageza Group analyst. "Then it will be able to figure out which projects can be done on what type of machine. The software juggles those demands and directs the flow of the work."

At present, truly global grids are only being used by scientists and academics who perform data-intensive analysis and experimentation. However, these programs are demonstrating jaw-dropping results that are sure to make an impact in the CIO's office.

The best known example is the Search for Extra-Terrestrial Intelligence (SETI), a project that harnesses unused PC power to analyze radio waves from space in the search for alien life. SETI organizers estimate they are receiving approximately 15 teraflops of processing power annually at a cost of $500,000. Compare that to IBM's ASCI White supercomputer, which delivers 12 teraflops of performance for $110 million, and you can see why a slew of new grid computing trade shows has sprouted up in the past two years.

While certainly not the only global grid initiative going, the SETI program makes it easy to see the business case for building a grid. By distributing computing resources, enterprises can cost-effectively manage internal IT investments while maximizing computing power and enabling users to perform complex computations and simulations that require computing power that few, if any, corporate data centers possess.

Early Adopters
Industries with intensive R&D efforts have been the earliest adopters of grid technology, specifically in aerospace, pharmaceuticals, biotechnology, semiconductor design and automotive. "Grid computing is good for large computations with small data transfer that don't require shared memory," said Manuel Peitsch, director of knowledge management for $41 billion pharmaceutical giant Novartis. For companies like these, building an internal grid makes economic sense because of the time- and processor-consuming experiments and simulations being run by their engineers and scientists. However, grid is not a one-size-fits-all solution.
"In time, deploying this type of task management software could become a real alternative to buying a lot of new servers and computers," said Pie Rieppo, an analyst with Gartner. "But right now, it's still mainly going to appeal to a fairly technical class of end users."
Estimates of the grid software market bolster the analyst assessment, as Gartner estimates the size of the grid computing market at no more than $150 million today. And while Gartner believes this market has the potential for steep and substantial growth of up to 50% as it matures, it could be more than 10 years before it reaches $1 billion.

Market Players
Since initial grid efforts were meant to help universities and research institutions replicate the performance of standalone mainframes and supercomputers, it should come as no surprise that the primary players in today's grid efforts are the big hardware players – HP, IBM and Sun – as well as up-and-coming software vendor Platform. Each of these players sees the opportunities for grid computing differently, and this perception is visible in their product initiatives.