Performance evaluation of multiprogrammed time-shared computer systems
Akira Sekino · 1973
A comprehensive set of hierarchically organized modular analytical models is developed for performance evaluation of multlprogralmned vlrtual-memory tlme-shared computer systems. This hierarchy of models includes a user behavior model, a secondary memory model, a progrembehavior model, a processor model, and a total system model. The thesis especially details on the last three models. The program behavior model permits estimation of the frequency of paging. The processor model evaluates the throughput of a given multi-processor multi-memory processing system under multiprcgramming. Finally, the total system model derives response time distribution of an entire computer system under study. Accuracy of performance prediction by these models is examined by comparing the predicted performance and the measured performance of the Multlcs system. These analyses are then applied to the optimization of computer systems and to the selection of the best performing configuration for a given budget. This framework of performance evaluation not only guides human intuition in understanding actual performance problems but presents reliable answers to quantitative performance questions about throughput and response time of actual computer systems.