Preview

PDF, English Download (542kB) | Terms of use

Abstract

Determining a system or component’s dependability invariably involves some kind of statistical analysis of a large number of tests of its behavior under typical usage conditions, regardless of the particular collection of attributes chosen to measure dependability. The number of factors that can affect the final figure is therefore quite large, and includes such things as the ordering of system operation invocations, the test cases (i.e. the parameter values and expected outcomes), the acceptability of different operation invocation results and the cumulative effect of the results over different usage scenarios. Quoting a single dependability number is therefore of little value without a clear presentation of the accompanying factors that generated it. Today, however, there is no compact or unified approach for representing this information in a way that makes it possible to judge dynamic systems and components for their dependability for particular applications. To address this problem, in this paper we describe a new, compact approach for presenting the tests used to determine a dynamic system’s dependability along with the statistical operations used to turn them into a single measure.