Cloud Provisioning for Work ow Application with Deadline using Discrete PSO

Abstract

The need of cloud consumers to optimize all options offered by cloud provider has been rapidly arisen during the recent years. The consideration involves the appropriate number of VMs must be purchased along with the allocation of supporting resources. Moreover, commercial clouds may have many different purchasing options. Finding optimal provisioning solutions is thus an NP-hard problem. Currently, there are many research works discussing the cloud provisioning cost optimization. However, most of the works mainly concerned with task scheduling. In this paper, we proposed a new framework where number of purchased instance, instance type, purchasing options, and task scheduling are considered within an optimization process. The Particle Swarm Optimization (PSO) technique is used to find the optimal solution. The initial results show a promising performance in both the perspectives of the total cost and fitness convergence. The designed system provides the solutions of purchasing options with optimum budget for any specified workflow-based application based on the required performance.