Grid computing is the term applied to the combination of computer resources from multiple administrative domains used to complete one common task. With grid computing, this task usually has an identifiable scientific, technical or business end aimed at solving a problem that requires an inordinate number of computer processing cycles and or the need to arrange and codify huge amounts of information.
One of the most important tactics employed in grid computing is using specialist software to divide and distribute individual pieces of a problem or program among several computers and with today’s grids, this can involve thousands of computers. The whole point of distributing individual tasks to each computer is that each computer will be able to crunch unbelievable amounts of data if it only has to work with one task.
Grid computing is therefore a shining example of distributed, large-scale cluster or super-computing. It pleases work-study engineers as it is a fine way of moving forward via network-distributed parallel processing, essentially the Holy Grail of efficiency and work-study.
The main advantage of grid computing is that the grids involved can be distributed between premises and even companies who have agreed to work on the same project or problem. Indeed, the thing that really distinguishes grid computing from more traditional cluster computing systems is that grids actually tend to be geographically dispersed. You could have the best coldfusion hosting you can get online however it may not provide the same business continuity or disaster recovery traits as a grid computing based platform like Amazon’s E32.
In terms of grid computing as an alternative to supercomputers, grid computing is, in general, a particular kind of parallel computing that requires many complete computers (‘complete’ meaning computers with fully functional onboard CPU, storage, power supply, network interface, etc) which are then connected to a network (private, public or internet) by a generic network interface, such as Ethernet. Therefore, it is very different to the conventional supercomputer, which runs many processors connected by a local high-speed computer bus. This generally rules out the possibility of connecting the processes of a supercomputer with the processes of any other outside agent and the bonuses of parallel processing are generally lost.
Grid computing is also cheaper to set up than a supercomputer, particularly if you can get more parties involved. This is mainly due to the facts related to grid computing hardware and the joys of buying in bulk. Supercomputers tend to run on purpose-built parts and processors whereas successful grid computing can achieve similar results to a supercomputer by using more generic parts as part of a predetermined parallel process.
The main disadvantage of using grid computing is that all results of all processes will have to be sent from place to place and collaboratively assessed in order for any final outcome to be deuced. This can be a strain in terms of timing issues as well as the fact that grid computing is heavily reliant on the ever-unpredictable notion of managing disperse data as well as connectivity. Data Management is a key consideration in today’s cloud based commuting environment.
Still in mid-2009 it was found by scholars between Cambridge, Oxford, Manchester, Kent and London that they could process excellent results which rivalled the supercomputers of MIT and Caltech. This was a major success for grid computing. So, grid computing is an excellent (though possibly underdeveloped) cheaper alternative to the use of supercomputers. Both configurations have their advantages and disadvantages. Supercomputers are great for crunching large amounts of data and passing out onsite results whereas the cheaper grid computing configurations requires connectivity in order for results to be combined and analysed properly.
This Grid Computing Portal seeks to foster collaborative development of grid technology, and related HPC, by assembling, exposing, and archiving grid technologies in a public forum. Our audience is everyone involved in grid technology, from developers and implementers to technology managers, research scientists, students, industry leaders, and beyond.
Our goals are to track, update, and summarize grid projects worldwide, and provide a current and ongoing information distribution channel for national and international grid projects, to include PI names and sponsors, academic and commercial affiliations, funding opportunities, descriptions of purpose and projected outcomes, reference implementations, middleware grid services, certificate authorities, authentication strategies, and news channels including alerts, conferences, and training opportunities, and a communications section for developers.