VM Provisioning Policies to Improve Profit and SLA Violation of Cloud Service Providers

Abstract

Cloud computing is an emerging technology in the IT world. Some features of cloud, such as low cost, scalability, robustness and availability are attracting large-scale industries as well as small business towards cloud. A virtual machine (VM) can be defined as a software that can run its own operating systems and applications like an operating system in physical computer. As the number of users increases, allocation of resources and scheduling become a complex task. The optimization of VM provisioning policies offer improvement like increasing provider's profit, energy savings and load balancing in large datacenters. In cloud computing when resource requirement of user's requests exceed resources limits of cloud provider, to fulfill the requests the cloud provider outsources to other cloud providers resources, this concept is known as cloud federation. In this paper we propose two algorithms for VM provisioning in federated cloud environment. The first approach tries to improve the cloud providers profit and the second, to minimize the Service Level Agreement (SLA) violation through load balancing. The proposed load balancing algorithm is threshold based. We consider two types of pricing models for VMs, on-demand and reserved instances of VM. Simulation results show that the proposed algorithm reduces the SLA violations.