The evaluation of core allocation algorithms in a large scale computer by a trace-driven model

In this thesis we present an analysis of the operating system of a large scale information system, specifically the Exec VIII, currently implemented on the Univac 1108 at the University of Houston. To accomplish this, we have developed a trace driven-model which uses run stream characteristics as taken from the accounting log maintained by the operating system as input. We present a description of the model and a comparison of significant characteristics of the operation of both the operating system and the trace driven model in order to show that the model correctly simulates the operations of the operating system under observation. With the valid model, we analyze various storage allocation algorithms and possible storage configuration changes to determine the effect of each on the efficiency of the operation of the operating system.