Grid Computing Not The Right Answer for High Performance Business Applications
I recently had a long conversation with a senior IT executive from a very large investment bank - the type with an annual IT budget measured in hundreds of millions and a small army of application developers. It is one of those rare organizations that actually uses IT for competitive advantage (whereas most companies are just happy for things to just work). When the topic moved on to application performance, he revealed internal research had found very few areas where it might be appropriate to use grid computing - the types of application that require massively parallel mathematical computations. However, initial experiments with projects revealed that grid computing turns out to add little value for typical business applications that involve lots of small steps and the occasional bottleneck. It also turns out that grid computing is very difficult to use with typical business applications to take advantage of mutli core processors. There is just a technology mismatch.
It seems word is spreading about the bad experiences with grid computing. A recent report by of IT executives by technology integration company Morse (a leading provider to the London financial services market) indicates that only eight percent of UK organizations have adopted or even planning to adopt grid computing. It goes on to say "opponents voiced concerns such as cost, complexity and security as the main reasons for not getting into Grid." Even worse, an amazing 88 percent "said their board of directors didn't feel that Grid computing was worth putting time and energy into."
So yet another technology that got hyped but they did not live up to expectations.
PJ Murray
CodeFutures Software