An Optimized Approach for Assisted Firewall Anomaly Resolution

The security configuration of firewalls is a complex task that is commonly performed manually by network administrators. As a consequence, among the rules composing firewall policies, they often introduce anomalies, which can be classified into sub-optimizations and conflicts, and which must be solved to allow the expected firewall behavior. The severity of this problem has been recently exacerbated by the increasing size and heterogeneity of next-generation computer networks. In this context, a main research challenge is the definition of approaches that may help the administrators in identifying and resolving the anomalies afflicting the policies they write. However, the strategies proposed in literature are fully automated, and thus potentially dangerous because the error-fixing process is not under human control. Therefore, this paper proposes an optimized approach to provide assisted firewall anomaly detection and resolution. This approach solves automatically only sub-optimizations, while it interacts with human users through explicit queries related to the resolution of conflicts, as their automatic resolution may lead to undesired configurations. The proposed approach also reduces the number of required interactions, with the aim to reduce the workload required to administrators, and employs satisfiability checking techniques to provide a correct-by-construction result. A framework implementing this methodology has been finally evaluated in use cases showcasing its applicability and optimization.