Implement of Dynamic Time Quantum Shortest Load First Scheduling for Efficient Load Balancing

Abstract

Cloud Computing is a modern eminent technology in this era. Cloud service providers requires handling vast request for scaling up the size of cloud, but Cloud Computing faces critical challenges on its glorious future. Those challenges create problem for exploring for its perfect realization. Load Balancing is one of the hindrances in that case. Load Balancing is a process of distributing workload among different others nodes of distributed system to improve the perfect utilization of resources and job response time. Load Balancing should be done properly, as failure in any one of the node can lead of unavailability of data. There are already different research works has been done on Load Balancing using various scheduling techniques such as Round Robin (RR), First-ComeFirst-Served (FCFS), Shortest Load First (SLF) etc for reduce the execution time, reduce average waiting time for giving better quality of services in Cloud Computing. In this paper new scheduling has been approached ,that is Dynamic Shortest Load First Algorithm (DSLF).Shortest Load first is one of the sophisticated scheduling algorithm, for reduce the average waiting time compare to other existing algorithm and give better load balancing but this has some drawbacks. The burst time of processes have to predict before CPU start the execution. But it is impossible to predict the burst time of the process before the process complete the execution. If the process did not end their execution in definite predicted burst time then which algorithm will execute, then this question will come. In case of FCFS, starvation will occur, though the average waiting time will increase in that case. In my approach the main concept is, if the process remain in ready queue after ending of their predicted Burst Time, a new quantum will create dynamically and each unfinished process will get the amount of time quantum to execute in CPU if any process does not completed after execution of that amount of time (Time Quantum) it will go to tail of the ready queue and waiting for its turn [1].