Abstract
IT infrastructure is a crucial part in most of today's business operations. High availability and reliability, and short response times to outages are essential. Thus a high amount of tool support and automation in risk management is desirable to decrease outages. We propose a new approach for calculating the root cause for an observed failure in an IT infrastructure. Our approach is based on Abduction in Markov Logic Networks. Abduction aims to find an explanation for a given observation in the light of some background knowledge. In failure diagnosis, the explanation corresponds to the root cause, the observation to the failure of a component, and the background knowledge to the dependency graph extended by potential risks. We apply a method to extend a Markov Logic Network in order to conduct abductive reasoning, which is not naturally supported in this formalism. Our approach exhibits a high amount of reusability and enables users without specific knowledge of a concrete infrastructure to gain viable insights in the case of an incident. We implemented the method in a tool and illustrate its suitability for root cause analysis by applying it to a sample scenario.
Introduction
Modern IT infrastructures are characterized by a complex web of dependencies. A single service, such as email, may rely on a chain of servers, databases, network switches, and power supplies. When a failure occurs, identifying the root cause is often difficult because the visible symptom (e.g., a service outage) is far removed from the underlying fault (e.g., a misconfigured network switch). Traditional Root Cause Analysis (RCA) tools often struggle to balance the rigid structural rules of IT architecture with the inherent uncertainty of hardware failures or cyber-attacks.
Methodology
The proposed framework utilizes Markov Logic Networks (MLNs) to unify logical reasoning with probabilistic modeling. An MLN consists of First-Order Logic formulas with associated weights. "Hard rules" represent absolute dependencies (e.g., if the power supply fails, the server fails), while "soft rules" represent probabilistic risks (e.g., the likelihood of a hard drive failure).
Abductive Reasoning
While standard MLNs are deductive—predicting effects from causes—RCA requires abductive reasoning to infer causes from observed effects. The authors adapt a method to "flip" the logic: for every rule $Cause \rightarrow Effect$, a reverse implication $Effect \rightarrow (Cause_1 \lor Cause_2 \dots)$ is added. To manage the complexity of multiple possible explanations, the authors introduce a "rare risks" optimization. By assigning negative weights to all risks, the system is biased toward finding the simplest and most probable explanation that accounts for all observed evidence. The implementation leverages the high-performance RockIt solver to perform Maximum A Posteriori (MAP) inference.
Case Studies and Results
The system was evaluated using real-world scenarios from a research group's infrastructure.
The Printer Scenario
In a scenario involving a multifunction printer, the system correctly identified an authentication issue (LDAP) as the root cause for a scanning failure, even though printing and copying services remained functional. This conclusion matched the findings of human experts after extensive manual investigation.
The SVN Server Outage
During an internal code repository outage, the system initially suggested a virtual machine (VM) overload. When provided with evidence that the VM was healthy, the system dynamically updated its inference to identify network congestion on the physical blade server, caused by a large data transfer on a separate VM. This iterative process highlights the "human-in-the-loop" potential of the tool.
Conclusion
This hybrid approach demonstrates the power of combining logic and probability for IT troubleshooting. By leveraging industry standards like the IT-Grundschutz Catalogues for risk weighting, the model remains grounded in practical security taxonomies. Future work will focus on automating the translation of infrastructure maps into logic formulas and testing the system's scalability in large-scale enterprise data centers.
Summary updated