Five Things To Know About Grid Computing

Grid computing. Every time I hear that term, I think of a checkerboard. But that analogy is a bit misleading.

Picture computers as checkers on a board. When you move a checker, the whole complexion of the board changes. That's not really so with grid computing, which IBM defines as "All or some of a group of computers, servers and storage across an enterprise, virtualized as one large computing system." Grid computing ("grid") is getting increased press these days, although the concept is not much different from an older term, "clustering." Not much different, but there are some important ways the two diverge. If you're not up to speed on this area, here are five things you should know about grid computing. Consider it a cheat sheet for when the boss asks you "What about this grid computing?" And with the cost economies grid can afford, you can be sure that day is coming.

1.) Grid computing is different from clustering.Clusters are computers spread out among an organization, and a central resource manager manages resource allocation. Not so with grid computing. PCs are spread out, but each node has its own resource manager. The point is not necessarily to have a single system view, but rather to harness the unused power of many different machines. Grid computing can be seen as an evolution from peer to peer as well as from clustering arrangements.

2.) You can marry Web services with the grid.Web services enable applications that live on different PCs to communicate, while grid computing finds unused resources on any number of networked computers to solve computational problems that too complex for one machine alone to handle. It's a union made in heaven!

3.) Grid computing can increase productivity, and therefore the bottom line.Clearly, setting up a grid can improve productivity, because end users have unlimited access to the computing resources they need, when they need them. Simply put, more people are able to get more things done. Cycle times are reduced and workloads are consolidated. Grids can be built using existing infrastructure, which keeps company expenditures low. Getting more back for the IT buck is the mantra at companies these days, and grid computing can play a major role toward that end (see "Slow But Steady Growth For 2004.")4.) Grid technology is not just for Intel-based machines. This one is for my Apple fans. Earlier this year, Apple introduced its Xgrid Technology. The technology is aimed at turning a Mac cluster into a supercomputer, which is the aim of all grid-computing schemes. With Xgrid running on Xserve G5 servers in a 42U rack, up to 84 Power PC G5 processors can be clustered to create a supercomputer with 1.5 teraflops of processing power. Mac users will say, "Whoa!" (PC users will flood me with e-mail claiming it's all hype.)

5.) The user of a desktop PC that is contributing processing power will experience no negative effects due to the connection. The point here is for all of this to be seamless. If a user needs to run an application that requires more horsepower, the work on the machine will automatically be reallocated to another machine within the grid.

A number of different industries are currently using grid technology, including financial services, academia (see "Harvard And IBM Build Grid and insurance (see "Is Grid Computing Next Wave?" . It's certainly a technology that is worth looking at, along with server consolidation and virtualization, to keep you company running lean and mean. Let me know what you think about grid computing, and whether you are using it where you work.