© 1986 by British Computer Society
| ||||||||||||||||||||||||||||||||||||||||||||||||||
System Design and Structuring
Computing Laboratory, University of Newcastle upon Tyne, Claremont Road, Newcastle upon Tyne, NE1 7RU, UK
The task of implementing a large and sophisticated computing system is often unduly costly and time-consuming, with the resulting system exhibiting inadequate performance and reliability, because of excessive system complexity. Such complexity can be reduced significantly by ensuring that the system is constructed out of a well-chosen set of largely independent components, which interact in well-understood ways. However, the task of structuring a system, i.e. of choosing and defining appropriate components, can be very difficult. This paper describes a technique of system structuring which involved distinguishing the functionality which a system is intended to have from other desirable attributes, such as reliability and security, and then using separate components to provide each of these attributes. Various UNIX-based systems which have been implemented at Newcastle are used to illustrate this structuring technique.
Received March 1985.
* Computing Laboratory, University of Newcastle upon Tyne, Claremont Road, Newcastle upon Tyne, NE1 7RU