Application-Level Checkpoint/Restart for Large-Scale Attack and Compliance Graphs

Attack graphs and compliance graphs are graphical representations of systems used to analyze their standings in regards to cybersecurity or compliance postures. These graphs are designed to examine all permutations and combinations of changes that can occur to systems, and represent those changes through the topological information of the resulting graph. Due to their exhaustive nature, these graphs are often large-scale. Various efforts have shown success in the parallelization and generation deployment on High-Performance Computing (HPC) systems. Despite the additional compute power that HPC systems provide, these large-scale graphs still posses lengthy runtimes and require large amounts of system memory to fully contain the resulting graph. This work presents a Checkpoint/Restart (C/R) implementation specific to attack and compliance graphs that aims to provide fault-tolerance and memory relief throughout the generation process. This implementation has been designed to be built within the graph generation source code, and provides a lightweight alternative to other C/R implementations. The results indicate that this implementation is successful at providing fault-tolerance and memory relief while having low impact to the generation process and to the source code.