Efficient Verification Of Network Fault Tolerance Via Counterexample-Guided Refinement

We show how to verify that large data center networks satisfy key properties such as all-pairs reachability under a bounded number of faults. To scale the analysis, we develop algorithms that identify network symmetries and compute small abstract networks from large concrete ones. Using counter-example guided abstraction refinement, we successively refine the computed abstractions until the given property may be verified. The soundness of our approach relies on a novel notion of network approximation: routing paths in the concrete network are not precisely simulated by those in the abstract network but are guaranteed to be "at least as good." We implement our algorithms in a tool called Origami and use them to verify reachability under faults for standard data center topologies. We find that Origami computes abstract networks with 1-3 orders of magnitude fewer edges, which makes it possible to verify large networks that are out of reach of existing techniques.