The term ‘Grid computing’ was coined during the 1990s to describe the sharing of computing resources – in particular storage and processing resources – that are located in different places. This allows users to get on with using the resources without having to buy or maintain the equipment for themselves. In fact, in the Grid computing model the user may have no awareness of where resources are located or who runs them.
Successful Grid computing requires a number of elements. Firstly the physical resources need to be in place in various sites. Many HPC sites are now co-operating by allowing their resources (or certain proportions of them) to be part of computing Grids.
Secondly there must be software that enables users to access the resources seamlessly, while providing the necessary security and accounting information (‘shared’ access does not necessarily mean ‘free’ access). This software is termed ‘middleware’ and provides a layer which enables the users’ applications to access the Grid resources.
EPCC has been involved in many Grid projects, including running test-beds for Grid middleware and its use by applications, and development of the Grid middleware itself.
Grid computing is not only useful for scientific researchers. Business use of Grid computing is growing because of the attraction of lower costs of ownership and the emergence of service-based models for delivery of IT functionality.