NEOD: Network Embedded On-line Disaster management framework for Software Defined Networking

Network management against security, reliability, and performance attacks is an integral part of building dependable, high-performance network services. In traditional networks, the management practices take mainly remote approaches to cope with the ossified network infrastructure. However, since disaster events that occur within the network should be inferred by the remote management system on the network edge, the problems are often accumulated and enlarged, and the diagnosis is delayed, inaccurate, unreliable, and not scalable. Although a few embedded approaches become available, they are costly and limited to the vendor specific applications. Software-Defined Networking (SDN) architecture has been recently proposed in order to enable flexible network control plane. However, it has focused mainly on traffic engineering, network virtualization, and off-line configuration management problems, and there is very little research on on-line management against network disaster. In this paper, we propose a globally deployable Network Embedded On-line Disaster (NEOD) management framework for SDNs. NEOD addresses important network management issues including agility, accuracy, reliability, and scalability. We have identified critical network disaster management metrics to adaptively support versatile application requests on SDNs. We have implemented the proposed system and metrics in OpenFlow with OpenWrt based routers, and have shown the effectiveness of NEOD through extensive system experiments as well as Mininet-based simulations.