Evaluating CPU Utilization in a Cloud Environment

Virtualization is used to ease computing resource management, resource utilization and running multiple heterogeneous or homogeneous operating systems on a single physical machine. Virtualization offers many advantages such as reducing fiscal costs, easing system migration, running legacy applications, easing backups and disaster recovery, and utilizing the shared resources over traditional data centers. In virtualized environments, the higher the percentage of the CPU utilization results in the maximum performance. However, virtualization systems have shortcomings including virtual machines allocation, virtual CPU configuration, and virtual CPU to physical CPU mapping. Such shortcomings may lead to system performance degradation. In this thesis, we focus on the role that hypervisors, virtual CPU to virtual machine allocation, and the virtual CPU to physical CPU mapping play on CPU utilization. PTS benchmarking tool is used as a trafic. We analyzed the implication of virtualization technolgoy layer, vCPU-VM assignment, and vCPU-pCPU mapping on performance in term of CPU utilization. PTS benchmarking tool is used since it can track CPU performance and it is also a useful data point tool for performance investigation. The results of this study will help cloud service providers and researchers to select and decide the best virtualition techerlogy, vCPU-VM configuration, and mapping strategies for their specific needs.