A Theoretical Analysis of System Quality

Specification of the rationale for developing a system quality measure entails considerable analysis and a careful evaluation of certain fundamental and often cherished reliability measures, e.g., mean time between failures or mean uptime. From such an analysis has come the following conclusions. In order to adequately assess the effects upon system performance of differing kinds and amounts of redundancy and to predict more accurately system availability, any nonclassical measure should quantify both length and quality of performance. A measure has been developed that meets such requirements, is distribution free, and can be evaluated in terms of system history. Two evaluational models with their peculiar difficulties are discussed.