Quantitative survivability evaluation of three virtual machine-based server architectures

Virtual machine-based services have become very popular in data centers and cloud computing in recent years. Efficient redundancy technology of virtual machine provides good availability for services; thus, it has become possible to provide continuous services even if the system suffers intrusions. So far, many intrusion tolerant server architectures have been developed based on virtual machine technology in academia and industry. Unfortunately, to the best of our knowledge, there is very little work done on evaluating the survivability of virtual machine-based server architectures in the literature. In this paper, we focus on analyzing and evaluating the survivability of three virtual machine-based architectures, which are load balance server architecture (LBSA), isolated component server architecture (ICSA), and Byzantine fault tolerant server architecture (BFTSA). As a reference, a traditional server cluster without virtual machines is also analyzed and compared. We model different architectures with Continuous Time Markov Chain (CTMC). We analyze the transient behaviors and steady states of different virtual machine-based architectures. Furthermore, the costs of the traditional server cluster and the virtual machine-based architectures are compared in terms of processing costs, memory costs, communication costs, and fail-safe fault tolerance. The results show that BFTSA has better survivability than LBSA and ICSA, but with longer time to reach the steady states and higher communication costs.