Towards optimal path encoding using SAT solver

Abstract

A Software-Defined Networking with programmable data-plane allows applications to realize policies by generating both datapath layouts and rules on switches. However, computing optimized datapath layouts and populating the compact rules for such datapaths can be overwhelmingly complex. This paper introduces a generic approach to realize network-wide policies in network, then derive optimal encoding for paths and forwarding rules for switches by modeling the process as a SAT problem. Finally, the evaluations on real ISP topologies show that our solution achieves about a 50% reduction of the path encoding length in average compare with the state of the art.