An Efficient Solution For Task Scheduling And Balancing Load In Cloud Using Hybrid Approaches

Abstract

ABSTRACT newline newlineThe cloud computing technology has brought a significant shift in the way the IT resources are handled for several applications. It is indeed the future of all distributed computing applications that would utilize on-demand computing resources. Users are increasingly outsourcing their demand for data to cloud resources. In spite of such advantages offered by cloud technology, the issue of load balancing plays a major role in achieving the required performance. Cloud performance can be improved through dynamic scheduling of the tasks to the available resources by keeping a watch on prevailing load on each cloud machines. In this research work, various hybridized algorithms have been proposed for obtaining an optimized solution for the task scheduling issue by balancing the available load. A multi-objective function has been formulated against which the proposed algorithms have been tested. Several QoS parameters in different combinations involving formulation of the multi-objective function have been included. Firstly, makespan and cost parameters were considered and a new hybrid algorithm named League Champion Whale Optimization Algorithm (LCWOA) had been proposed for improving the performance of cloud by reducing the cost and makespan parameters. The LCWOA is the hybridized version of the Whale Optimization Algorithm and the Hybrid League Algorithm. The multi-objective function has been designed taking into account the cost of the CPU and memory consumed in completing the tasks received and accordingly a fitness value has been estimated. newline newline newline newlineThe proposed LCWOA showed a considerable improvement with respect to makespan and cost parameters, over a set of the existing algorithms like PSO, Firefly, and GA. Secondly, the research work focuses upon balancing the load in the cloud environment where independent tasks are allocated to the corresponding resource through use of a Modified Canopy Fuzzy C-means Algorithm (MCFCMA). Tasks were allocated to the corresp