Published Paper Details:

Job scheduling is state of the art problem in advanced computing system. To tackle the issue of larger Makespan and Flowtime, several techniques are being researched over. This paper works towards creation of job scheduling policy where burst time is considered for arranging the jobs in clusters. Proposed system is categorised into two phases: first phase arranges the jobs by following shortest job first scheduling. The queue thus formed is presented to round robin scheduler with time quantum that varies depending upon the burst time of job. Jobs arranged are arranged in batches of 10% of total jobs in queue. SJF scheduler considered is non premptive where RRS scheduler is premptive. Second phase executes the jobs by looking at the resource clusters. Multi-source shortest path dynamic algorithm is used for selection of job that can be assigned to the resource cluster. Once job execution is complete credits are assigned which will be from 0-10. Higher the credit more proficient is the result. Optimal result is obtained by the application of proposed system in terms of Makespan and flowtime. Simulation is conducted in MATLAB showing improvement of 6% in overall result.