Optimal scheduling for multi-flow update in Software-Defined Networks

摘要

Aiming to adapt various traffic dynamics, deal with network errors, perform planned maintenance, etc., flow update is carried out frequently in Software-Defined Networks (SDNs) to adjust network configurations, and how to update the flows successfully is an important and challenging problem. Due to the resource constraints of network links and switches, the update of multiple flows can cause transient congestion which degrades network performance if they are not scheduled carefully. In this work, we address the multi-flow update sequence scheduling problem by formulating it as an optimization problem. Considering the limitations of link bandwidth and flow table size, we find the optimal flow update sequence with efficient resource utilization. By extensive simulations under real network settings, we demonstrate the effectiveness of our algorithm, which achieves smaller flow table usage, and increases the update success rate by up to 20% without sacrifice of link utilization. (C) 2015 Elsevier Ltd. All rights reserved.