Evaluations of Network Performance Enhancement on Cloud-native Network Function

Abstract

Cloud-native network function (CNF) has progressively become a considerable solution adopted by telecom operators to provide network services. The deployment architecture of CNF can support a massive scale with low operational overhead compared to virtual network function (VNF). However, the network performance in CNF encounters severe challenges. One of the hurdles is Packet processing delay, whether in the user space or kernel space. In CNF, both spaces have solutions intend to accelerate the efficiency of processing. However, CNF is an emerging architecture with a lack of evaluation of which technology we should adopt. In this research, we experiment with the different processing acceleration technologies along with different CNF deployment methods. In our experiments, we believe that the kernel space has more opportunity for network acceleration in CNF, especially when CNFs deploy in virtual machines (VM).