Abstract: Load balancing and Task scheduling are the two major research challenges in cloud computing environments. Aiming at these challenges, this paper proposed a new mechanism called as Heuristic Clustering and Multi-Objective Task Scheduling (HC-MOTS). HC aims at load balancing over physical hosts which have totally dynamic resource availability. From the available hosts in cloud data center, HC picks up only optimal set of physical hosts by considering their dynamic characteristics like Posterior Probability, remaining CPU resource and remaining Memory resource. Further, the task scheduling is modeled as multi-objective fitness function formulated based on three objective functions with Execution time, Cost and Resource utilization. For each function, an individual weight is assigned and they are iteratively optimized to get an optimal solution. Simulation experiments have been conducted on datasets namely GOCJ and Synthetic Workload dataset and the performance is assessed through Makespan and Average Throughput. Further, the comparison proves the superiority of proposed approach over state-of-the-art methods.

Cloud computing, Execution time, Load balancing, task scheduling, CPU, Memory.