Analysis of VM migration scheduling as moving target defense against insider attacks

Abstract

As cybersecurity threats evolve, cloud computing defenses must adapt to face new challenges. Unfortunately, due to resource sharing, cloud computing platforms open the door for insider attacks, which consist of malicious actions from cloud authorized users (e.g., clients of an Infrastructure-as-a-Service (IaaS) cloud) targeting the co-hosted users or the underlying provider environment. Virtual machine (VM) migration is a Moving Target Defense (MTD) technique to mitigate insider attacks effects, as it provides VMs positioning manageability. However, there is a clear demand for studies quantifying the security benefits of VM migration-based MTD considering different system architecture configurations. This paper tries to fill such a gap by presenting a Stochastic Reward Net model for the security evaluation of a VM migration-based MTD. The security metric of interest is the probability of attack success. We consider multiple architectures, ranging from one physical machine pool (without MTD) up to four physical machine pools. The evaluation also considers the unavailability due to VM migration. The key contributions are i) a set of results highlighting the probability of insider attacks success over time in different architectures and VM migration schedules, and ii) suggestions for selecting VMs as candidates for MTD deployment based on the tolerance levels of the attack success probability. The results are validated against simulation results to confirm the accuracy of the model.