Parametric Optimization Of Some Critical Operating System Functions An Alernative Approach To The Study Of Operating Systems Design
Abhilasha Tibrewal, Tarek Sobh · 2020
Operating systems theory primarily concentrates on the optimal use of computing resources.The study of operating systems design and concepts by way of parametrically optimizing critical operating system functions is the focus of this work.Our work marks a new approach to teaching and studying operating systems design processes.The four specific operating system functions studied are those of CPU scheduling, memory management, deadlock/synchronization primitives and disc scheduling.The aim of the study is to first introduce and discuss the modules in light of previous research, discuss in details the affecting parameters and their interaction and attempt to optimize some of the lesser-established parameter-performance relationships by way of simulations.Results of the simulations of the four functions are then analyzed in light of specific parameters and the effect they have on the overall system performance.System performance is judged by many measures, including: average turn around time, average waiting time, throughput, CPU utilization, fragmentation, response time, and several other module specific performance measures.Some of the parameters studied in the CPU scheduling module include: the round robin time slot, aging parameters, preemption switches and context switching time.Simulation of multilevel feedback queues is attempted and the performance is judged in terms of the above mentioned performance measures.In the context of memory management, some of the parameters studied include: memory size, RAM and disc access times, compaction thresholds, memory placement algorithm choice, page size and the time quantum value.The attempted simulation uses the continuous memory scheme.In the deadlock/synchronization module, the parameters studied include: the total number of processes, the total number of available resources and the maximum number of resources required by the processes.Four deadlock handling mechanisms are discussed and a deadlock avoidance algorithm is simulated.The denial (rejection) rate of requests for resources quantifies system performance.Within the discscheduling module, the parameters studied include: disc configuration/size, disc access time, disc scheduling algorithm choice, disc writing mechanism and all the parameters utilized in the memory management module.Performance is judged in terms of the above mentioned performance parameters and also the percentage seek and latency times.Some of the simulation specific results tend to highlight the role of optimizing the value of the round robin quantum in the modules, the importance of average seek and average latency times versus the system performance and the comparative performance of the various memory placement algorithms and disc scheduling algorithms.Lastly, an attempt to integrate the four specified modules is discussed to attain the final goal of designing an optimal operating system with the right permutation of design parameters to achieve excellent performance measures for various process mixes.