The static nature of current IT systems gives attackers the extremely valuable advantage of time, as adversaries can take their time and plan attacks at their leisure. Although cloud infrastructures have increased the automation options for managing IT systems, the introduction of Moving Target Defense (MTD) techniques at the entire IT system level is still very challenging. The core idea of MTD is to make a system change proactively as a means to eliminating the asymmetric advantage the attacker has on time. However, due to the number and complexity of dependencies between IT system components, it is not trivial to introduce proactive changes without breaking the system or severely impacting its performance. In this paper, we present an MTD platform for Cloud-Based IT Systems (MTD CBITS), evaluate its practicality, and perform a detailed analysis of its security benefits. To the best of our knowledge MTD CBITS is the first MTD platform that leverages the advantages of a cloud-automation framework (ANCOR) that captures an IT system's setup parameters and dependencies using a high-level abstraction. This allows our platform to make automated changes to the IT system, in particular, to replace running components of the system with fresh new instances. To evaluate MTD CBITS' practicality, we present a series of experiments that show negligible (statistically non-significant) performance impacts. To evaluate effectiveness, we analyze the costs and security benefits of MTD CBITS using a practical attack window model and show how a system managed using MTD CBITS will increase attack difficulty.
展开▼
