Conventional software availability models often assume only up and down state for the time-depdendent behavior of a software system. In this paper, we develop a planusible software availability model considering the degreneration of system performance in operation. We assume that the software system has two operational states during the operation phases: one is providing with full performance and the other is with degeenerated performance. The time-dependent behavior of the software system, which alternates between the operational and restoration state, is described by a Markov process. Several quantitative availability/performance measures are deived form this model. Especially, this model can derive computation software availability which is defind as the expected value of possible processing quantity per unit time at a specified time point. This is a measure taking account of reliability and performance simultaneously. Finally, numerical examples are presented for illustration of software availability/performance measurements and assessment.
展开▼
