Innovation and investment in IT tend to go in two waves. Innovation is driven by the speed at which IT companies can develop technologies and bring them to market at the right price. Investment in IT, on the other hand, tracks broader economic cycles, rising and falling in line with expansions or contractions in business activity.
While these two waves are often out of sync with one another, at present they have both hit a trough at the same time. Not only is the rate of growth in IT spending at a low point, but there are few new big ideas in IT. There is one exception, however, and that is a technology called grid computing.
Originally conceived as a way of harnessing spare computer capacity to carry out lengthy scientific calculations, grid computing is now being touted by IT suppliers such as IBM, Hewlett-Packard and Oracle as a way of making more efficient use of commercial IT resources and of improving the degree of collaboration that is possible between companies, particularly in the supply chain.
Searching for intelligence
Until recently, the best known application of grid computing was in an international project called the Search for Extraterrestrial Intelligence (SETI). In the quest to find signs of intelligent life, recordings of radio signals from the cosmos are downloaded for analysis by hundreds of home computers connected together in a grid network.
This collaborative exercise is co-ordinated by central software which makes use of idle periods on computers that are part of SETI@home to process the vast amount of data gathered by SETI scientists. Users do not have to worry about when the SETI program kicks in on their systems: this is controlled by a special screensaver. It’s an elegant solution to the problem of finding enough computing power for lengthy but repetitive number crunching.
The general idea of tapping into unused resources has now been adopted for more down to earth applications involving the servers, networks, databases and associated disk stores used in business computing. In some systems, particularly those that have been recently acquired, as much as 80 per cent of capacity can be unused.
Grid computing aims to let companies pool those resources, making better use of available capacity by spreading workload across systems.
Qualcomm, a US technology company that makes chips and software for third generation telecommunications systems, is using the grid computing model to run its supply chain management systems. The company says it has significantly improved the performance of its global planning and manufacturing systems and has sharpened the accuracy of its forecasting so that it can respond in near real time to peaks and troughs in demand.
‘We have a 97 per cent on time delivery accuracy rating. We don’t want to lose that – we want to get it better by improving the degree of control we have over the process,’ explains Norm Fjeldham, senior vice president and CIO at Qualcomm.
One of the advantages of the networked grid approach to computing is that it can be used to integrate systems run by different companies who want to collaborate. Qualcomm, for example, has set up a collaborative planning system with its suppliers. The system feeds into Qualcomm’s supply chain planning, exchanging demand forecasts and supply commitments as well as generating purchase orders and schedules.
With the aid of Oracle’s 10g grid database, one of the first commercial systems to use the grid computing model, Qualcomm sends its suppliers 20 weeks-worth of planning data in weekly batches. In this way everyone sees the same data and there is no need to exchange cumbersome spreadsheets.
‘For Qualcomm, grids are going to be a big part of our infrastructure,’ says Fjeldham. ‘We like the idea of being able to dynamically allocate resources. With grid we can move our resources to where they are needed and keep our total cost of ownership down.’
Whatever the cost benefits of grid computing, one thing is certain – it is a concept that chimes in well with a general trend to treat IT more as a utility than a set of separate solutions. Grid computing may not find intelligent life in the heavens but it may make our daily round on earth a bit smarter.