Security Solution on KVM Hypervisor for Detecting DoS Attacks on Cloud Servers

Abstract

In today’s time, virtualization is used by Cloud providers to provide multiple virtual machines to the customers. Even though there are various benefits of virtualization in terms of cost or maintenance, still there is a high possibility of attacks on cloud servers and hypervisors. In order to provide system-level protection to all the servers running under a hypervisor within the control of a hypervisor, implementing an intrusion detection system at the hypervisor level is highly recommended. While devising such a detection model, it is observed that in addition to the network layer and transport layer features of a network packet, the frame layer features are also having a high impact on the detection performance. This paper focuses on designing a hypervisor-based intrusion detection system that considers the frame layer features and analyzes their impact on intrusion detection performance. The detection model also uses three types of machine learning classifiers and the best classifier is identified out of our experimentation.