A simulator based performance analysis of multilevel feedback queue scheduling

Multilevel Feedback Queue (MFQ) algorithm allows the processes to switch between the queues depending on their burst time. For which the processes which are in ready queue come to first queue using Round Robin (RR). It comes to next queue when burst time is greater than quantum. Queues may have their own scheduling policy (such as RR, Shortest Job First (SJF) or First Come First Serve (FCFS)). In this paper, through a simulated MFQ program, we have shown that using RR in different queues and applying SJF for shortest job selection prior to RR from second queue onwards may improve the CPU utilization using the dynamically generated time quantum. The experiments using a set of processes have been done using MFQ simulator developed by us during the process. To run the simulator, CPU burst of a set of processes are to be entered and while doing so the system ignores the nature (CPU or I/O Bound) of each process. The dynamic time quantum for each queue is generated automatically by the simulator. Through a number of experiments performed, we observed that the performance of MFQ improves by applying SJF selection prior to RR algorithm from second queue onwards using dynamically generated time quantum as compared to other algorithms and for static quantum.