IT Risk Management with Markov Logic Networks

We present a solution for modeling the dependencies of an IT infrastructure and determine the availability of components and services therein using Markov logic networks (MLN). MLNs offer a single representation of probability and first-order logic and are well suited to model dependencies and threats. We identify different kinds of dependency and show how they can be translated into an MLN. The MLN infrastructure model allows us to use marginal inference to predict the availability of IT infrastructure components and services. We demonstrate that our solution is well suited for supporting IT Risk management by analyzing the impact of threats and comparing risk mitigation efforts.