SLA-Driven Dynamic Resource Management System for Private Clouds

Resource management in private clouds is a more challenging task than in public clouds. Because there are only finite resources in private cloud compare to public cloud with vast resources. Often, marginal resources are assigned to the application in the private cloud, which causes to the changes of service level of application executions. A resource management system is responsible for fulfilling SLA requirement so that it assigns resources when the allocated resource is not enough. Compare to conventional distributed environment, it is easier to meet the SLA requirement in virtualized one because there are many helpful mechanisms like consolidation, virtual machine migration, and dynamic scaling. We propose a resource management system for private cloud based on a novel server consolidation algorithm that uses live migration of the virtual machine, with dynamic scaling capabilities. We present our proposed architecture, the consolidation algorithm and the profile scheme to support dynamic scaling. The empirical experiments results show that our proposed approach outperforms other popular approaches in process time and load balancing.