摘要
Cloud computing is one of the most attractive cost effective technologies for provisioning information technology (IT) resources to common IT consumers. These resources are provided as service through internet in pay per usage manner, which are mainly classified into application, platform and infrastructure. Cloud provides its services through data centers that possess high configuration servers. The conservation of data centers energy give benefits to both cloud providers and consumers in terms of service time and cost. One of the fundamental services of cloud is infrastructure as a service that provides virtual machines (VMs) as a computing resource to consumers. The VMs are created in data center servers as the machine instances, which could work as a dedicated computer system for consumers. As cloud provides the feature of elasticity, the consumers can change their resource demand during service. This characteristics leads VMs migration is unavoidable in cloud environment. The increased down time of VMs in migration affects the efficiency of cloud service. The minimization of VMs migration reduces the processing time that ultimately saves the energy of data centers. The proposed methodology in this work utilizes genetically weight optimized artificial neural network to predict the near future availability of data center servers. Based on the future availability of resources the VMs management activities are performed. The implementation results demonstrated that the proposed methodology significantly reduces the processing time of data centers and the response time of customer applications by minimizing VMs migration.