A Parallelized Subgraphing Approach to Large-Scale Attack and Compliance Generation

The generation of large-scale graphs encounters unique challenges compared to the exploration of the same large-scale graphs. Attack graphs and compliance graphs can be useful applications of large-scale graphs in order for organizations to understand and analyze their cybersecurity or regulatory standings, but their generation can often be prohibitive. This work introduces a data-parallel distributed approach of large-scale graph generation with the specific application of attack and compliance graphs. This approach aims to distribute the exploratory states of a graph across compute nodes in order to compile groups of subgraphs that will compose the full graph. This work utilizes Message-Passing Interface as the standard for passing messages across nodes in the distributed generator program. Additional experimentation, results, and discussions are also provided regarding the viability of using a Distributed Hash Table. This work presents a first step, naïve approach to subgraphing, and discusses potentially viable alternatives for increased performance.