The influence of software structure on reliability

This paper assumes software structure to be characterized by the interfaces between subsystems or modules. Reliability is considered to be a measure of the extent to which the system can be expected to deliver usable services when those services are demanded. It is argued that reliability and correctness (in the sense used in current computer literature) are not synonyms. The differences suggest techniques by which the reliability of software can be improved even while the production of correct software remains beyond our reach. In general, the techniques involve considering certain unpleasant facts of life at an early stage in the design process, the stage where the structure is determined, rather than later. An appendix gives some specific examples of questions which, if they are thoughtfully considered early in the design, can lead to more reliable systems.