Abstract:
In the Cloud, computing resources need to be allocated and scheduled in a way that providers can achieve high resource utilization and users can meet their applications’ performance requirements with minimum expenditure. Due to these different intentions, there is the need to develop a scheduling algorithm to outperform appropriate allocation of tasks on resources. This research focuses on the resource optimization using a threshold-based tournament selection probability for virtual machines used in the execution of tasks. The proposed approach was designed in the creation of meta-task and a Median-Based improved Max-Min algorithm was used to process the jobs that went back to queue as meta-task. The research was able to allocate tasks to virtual machines efficiently thereby assigning tasks with large file size to machines with
maximum speed. The experimental results showed that the algorithm has better performance in terms of make span, resources utilization, and throughput. The load balance of tasks was also fairly distributed on the two Data centres and from the comparison of the different algorithms the bar chart showed that the throughput was also minimized.
