Recommendation of Hypervisors for System Resources Intensive Application Workloads based on their Performance in the Private Cloud

Abstract

Hypervisors enable cloud computing model to provide scalable infrastructures and on-demand access to computing resources as they support multiple operating systems to run on one physical server concurrently. This mechanism enhances utilization of physical server thus reduces server count in the datacenter hence driving the benefits of reduced IT infrastructure setup and maintenance costs along with power savings. The paper evaluates the system resources (CPU, Memory, Network, and Disk I/O) performance of three hypervisors, namely ESXi, XenServer and KVM in a CloudStack environment using the SIGAR-Framework for consolidated application workloads with concurrently running virtual machines. Based on the results, the paper recommends the hypervisors suitability for system resources intensive applications. The three hypervisors are carefully chosen to represent three categories (full virtualized, para-virtualized, and hybrid virtualized). We have created a private cloud using open source cloud computing software CloudStack. Hypervisors are deployed as hosts in the CloudStack in the respective clusters. Consolidated workloads are generated using open source load generator tools on multiple virtual machines of hypervisors and important system resources information is gathered using SIGAR framework.