Abstract:

In the virtual data-center, how to map virtual machines (VMs) to physical machines (PMs) is becoming a hot issue. However, most of existing VM scheduling schemes have not fully considered the reliability and dynamical workloads of VMs. This paper presents a novel bionic autonomic nervous system (BANS) based approach for cloud resource management. This approach supports self-optimization that provides a dynamic and autonomic way to adapt to dynamical workloads and VM resource requirements. For the VM allocation in the self-optimization, this paper presents a reliability-performance-energy correlation model that can model, analyze and evaluate reliability, performance and power consumption simultaneously.