Centralised Information Processing
In the very beginning, computers were large and expensive. In an organisation, there was only one computer and all processing took place in the "computing center". When a job needed to be done, the data was brought to the computing center, it was processed and the results of the processing then sent back to those who requested the job.
With the introduction of the concept of networking, terminals were added to the mainframe and so the data entry and the production of results could then be achieved at the place of origin. However, these terminals were dumb (had no processing capabilities) and the task of processing the information was still performed on the one centralised computer.
It terms of the totality of the task of working with and processing of information, decentralisation was taking place.
Many systems today still work using these principles. Example of these include Library systems and inventory control systems, especially those using the UNIX and PICKS operating system, and many mainframe environments.
With the introduction of microcomputers, the concept of centralised processing was considered by many to be an inappropriate way of processing data. However, with the introduction of Winframe technologies, Windows Terminal Server and Metaframe technologies, people are starting to reconsider the concept of centralised processing as a method to cut the cost of computing. The process simply put is to have a very powerful server that can run many instances of one or more applications and the user access this by using a simple "dumb" terminal. The savings are in the Total Cost of Ownership of the entire system.
Distributed Information Processing
With the introduction of smart terminals (such as Microcomputers) and Application Servers (such as NT Servers), the processing started to be done on the local computer. Now, when data needs to be processed, the user requests the application form the server and the data is entered and processed locally using the computing power of the local microcomputer on the desktop.
One step further to this is the concept of Client/Server. The philosophy behind this is that all computers are not equal and some computers perform some tasks better than others. So, when a particular task needs to be performed, the client connects via the network to the best computer for the job at hand, passes the data to that computer via the network to that computer which then carries out the processing and returns to the results to the originator.
At the same time, a second person may be doing a similar thing, only this time using the computer that the first person is using as terminal to do the processing required for that persons task. This is very typical of the environment where Silicon Graphics workstations are used such as at Curtin University.
Another way of distributing processing is the process of pipelining; this is particularily useful when a large amount of very similar data must have many different things done to it, such as in image processing. Each computer on the network is given one task to do and the data is sent to each computer in turn as a stream. Each computer does it's small task to each data element so at the end, all of the data is processed, but each computer has shared the task.
Another example of this is the processing of the data for the Search for Extra Terrestrial Intelligence (SETI) program. Data from the Aracebo Radio telescope in South America is distributed via the Internet to thousands of computer around the world. While these computers are not being used for their normal task, they process the data for SETI and return the results when required via the Internet.
Further information can be found in Computers: Tools for an Information Age, Capron, HL. P160/161
Mike Leishman