Modified Flower Pollination Based Task Scheduling in Cloud Environment using Virtual Machine Migration

Abstract

Cloud computing allows users to use resources pay per use model by the help of internet. Users are able to do computation dynamically from different location by using internet resources. The major challenging task in cloud computing is efficient selection of resources for the tasks submitted by users. A number of heuristics and meta-heuristics algorithms are designed by different researchers. The most critical phase is the selection of appropriate resource and its management. The selection of resource include to identify list of authenticated available resources in the cloud for job submission and to choose the best resource. The best resource selection is done by the analysis of several factors like expected time to execute a task by user, access restriction to resources, and expected cost to use resources. In this paper, cloud architecture for resource selection is proposed which combines these factors and make the effective resource selection. In this paper a modified flower pollination algorithm is proposed to migrate the task on efficient virtual machine. The selection of the efficient virtual machine is calculated by the fitness function. By calculating the fitness function, the modified FPA algorithm is used to take the decision regarding VM migration is required to improve the resource efficiency or not. In this paper Virtual machine mapper maps the task as per knowledge base i.e. past history of the virtual machine, task type whether computational or communicational based. The results are compared with the existing meta-heuristic algorithms.