Degradable computer systems with dependent subsystems

When building a model for degradable computer systems, it is not sufficient to merely quantify reliability and performance measures. These indices must be mathematically sound if they are to be used to design such systems in an optimal way. The paper presents an analysis of design optimisation for degradable computer systems and shows how this particular application leads to a system model with interdepedent subsystems. A procedure is presented on how to solve the resulting Markov model. Its computational complexity is compared to another solution method and shown to be largely more efficient.