Performance Evaluation of Hypervisors in the Private Cloud based on System Information using SIGAR Framework and for System Workloads using Passmark

Hypervisors using virtualization technology enable multiple operating systems to run on one physical server. Cloud computing model is less expensive because it streamlines the delivery of services by providing a stage for optimizing complex IT resources in a scalable manner with the help of virtualization technology and hypervisors. Selecting a suitable hypervisor for their organization’s private cloud is herculean task for modern CIOs. Hypervisor vendors do claim that they have negated virtualization overhead completely compare to native system, but still there exists minute virtualization overhead because virtual machines have to communicate with middle-layer hypervisor to access the underlying physical hardware and also there is an impact of other virtual machines running on the same hypervisor. Hypervisors are developed using different virtualization techniques like full virtualization, para-virtualization and hybrid model virtualization. This paper evaluates the performance of three hypervisors ESXi, XenServer and KVM using SIGAR framework for system information and Passmark for system workloads in the private cloud environment. Private cloud has been designed using open source cloud computing software CloudStack. Hypervisors are deployed as hosts in the CloudStack. This paper recommends best suited hypervisors for respective workloads in the private cloud based on the performance of system information and system workloads.