Abstract :
This paper presents an approach to system reliability involving s-dependence of the workload as well as the system configuration. Four classes of failures are described and then incorporated into the workload model. Mean time to failure and the system reliability are the functions of parameters estimated by monitoring a real system. The model allows multiple classes of users and priority requests to be represented. The model is validated using measurement data collected in an IBM installation.
Keywords :
"Hardware","Condition monitoring","Mathematical model","Integrated circuit modeling","Integrated circuit reliability","Circuit faults","Mathematical analysis","Reliability theory","Parameter estimation","Distributed computing"
