Inter-Flow Consistency: Novel SDN Update Abstraction for Supporting Inter-Flow Constraints

Software Defined Networks (SDN) are opening a new era in the world of networking by decoupling the data plane and control plane. With the centralized control plane, updating the networks becomes much more convenient to the network operators. However, due to the distributed nature of the data plane, people fail to avoid transitional states of SDN during network updates. The transitional states may be a combination of the old and the new network configurations, which may lead to incorrectness in forwarding behaviors and security vulnerabilities. This thesis complements the large body of consistent update mechanisms of SDN by proposing a novel network update abstraction, inter-flow consistency, which can guarantee certain relationships and constraints among different flows during network updates. To the best of our knowledge, we are the first to study the update consistency abstraction across different flows. We propose an update scheduling algorithm based on dependency graphs, a data structure revealing dependency among different update operations and network elements, in order to guarantee two basic inter-flow consistency, spatial isolation and version isolation. Also, we implement a prototype system with a Mininet OpenFlow network and Ryu SDN controller to evaluate the performance of our approach.