How Practical Are Intrusion-Tolerant Distributed Systems?

Abstract

Building secure, inviolable systems using traditional mechanisms is becoming increasingly an unattainable goal. The recognition of this fact has fostered the interest in alternative approaches to security such as intrusion tolerance, which applies fault tolerance concepts and techniques to security problems. Albeit this area is quite promising, intrusion-tolerant distributed systems typically rely on the assumption that the system components fail or are compromised independently. This is a strong assumption that has been repeatedly questioned. In this paper we discuss how this assumption can be implemented in practice using diversity of system components. We present a taxonomy of axes of diversity and discuss how they provide failure independence. Furthermore, we provide a practical example of an intrusion-tolerant system built using diversity.