Performance Assessment of Network Virtualization Technologies in Cloud Computing Environments: A Comparative Study of Leading Hypervisors

Abstract

Cloud computing represents a transformation paradigm in contemporary computer science applications. The foundation of cloud infrastructure relies heavily on virtualization techniques, whereby singular physical resources are transformed into multiple virtual counterparts. The virtualization process encompasses various components, including storage systems, computational processors, volatile memory units, networking infrastructure, and additional hardware resources. Both processor and network virtualization technologies deliver substantial benefits, including reduced hardware expenditures, decreased energy consumption, and minimized administrative overhead for resource management. This research concentrates on evaluating and analyzing network virtualization implementations across various cloud computing hypervisors while comparing their performance metrics. The comparative analysis is grounded in experimental studies conducted on three prominent platforms. Our findings demonstrate that virtual network architectures provide enhanced scalability and security compared to traditional physical networks. The capability to assign multiple virtual interfaces to individual virtual machines presents significant operational advantages. Through comprehensive experimentation comparing virtualized against non-virtualized environments across all evaluated hypervisors, we determined that virtual networks function effectively both with and without VLAN tagging mechanisms. Hyper-V exhibits superior flexibility and configuration options regarding VLAN tagging compared to XenServer or vSphere implementations.