Efficient Loop-Free Rerouting of Multiple SDN Flows

Abstract

Computer networks such as the Internet or datacenter networks have become a crucial infrastructure for many critical services. Accordingly, it is important that such networks preserve the correctness criteria, even during transitions from one correct configuration to a new correct configuration. This paper initiates the study of how to simultaneously update, i.e., reroute multiple policies (i.e., flows ) in a software-defined network in a transiently consistent and efficient manner. In particular, we consider the problem of minimizing the number of controller–switch interactions, henceforth called touches , while preserving fundamental properties, in particular loop freedom, at any time. Indeed, we empirically show that the number of such interactions affects the resource consumption at the switches. Our main result is a negative one: we rigorously prove that jointly optimizing multiple route updates in a consistent and efficient manner is $\mathcal {NP}$ -hard, already for two routing policies. However, we also present an efficient polynomial-time algorithm that, given a fixed number of correct update schedules for independent policies, computes an optimal global schedule with minimal touches. This algorithm applies to any per-flow independent consistency property, not only loop freedom.