Functional vs. Non-functional Properties

Abstract

THE SYN tools used throughout this monograph measure (for components) and predict (for systems) two behaviors: output (usually called ‘functional behavior’) and run time (an example of a contrasting ‘non-functional behavior’). As described in Chapter 14, the tools actually process as ‘run time’ an arbitrary secondary output from components and systems. The ‘non-functional’ behavior could be any property of interest, so long as there is a way to measure it at the component level and then to combine its component values into system values. For run time, measurement is straightforward using operating-system services and the series combination operator is addition, although concurrency introduces a bit of complication (Chapter 11). Throughout this chapter, run time continues to stand in for an arbitrary non-functional property. Reliability, which can be treated in much the same way as run time, is not at all straightforward and has its own Chapter 12.