Abstract

Software-Defined Networking (SDN) enables fast reconfiguration of the network based on the decisions of a logically centralized controller. While SDN is in transition from laboratories and greenfield data-centers to enterprise networks, the need for managing partial SDN deployments arises. This incremental deployment of SDN has its own set of challenges, for example the design of the mixed control plane or the integration in existing network management systems. In this paper we focus on the different speed of reconfiguration between legacy and SDN devices and how those differences constrain the overall reconfig-urability of the network. At first we present measurement results highlighting the differences between SDN and legacy devices in terms of reconfiguration speed. The measurements show an up to hundredfold difference of reconfiguration times between SDN and legacy devices. Second, we introduce a methodology which allows to evaluate the impact of these differences on the maximum reconfiguration rate of the whole network for different SDN deployment stages. Third, we investigate a large number of real world topologies and the gain in terms of reconfigurability that can be achieved by deploying SDN incrementally. The results show that even a small number of slow devices can severely constrain the maximum reconfiguration rate. Furthermore, even with the majority of the network nodes being replaced by SDN devices, the maximum reconfiguration rate increases in the best case only up to 5 times compared to the all-legacy network. Based on the findings we conclude that research has to focus more on the management of partial SDN deployments w.r.t. potentially large differences in reconfiguration times of the devices. Otherwise the potential of SDN lies dormant until the whole network is migrated to SDN-enabled devices.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call