In OpenFlow, control and data plane are decoupled from switches/routers. Direct programming of routers/switches is realised from one or more servers (so called controllers). In the case of an in-band OpenFlow network, the control traffic (traffic to or from the controllers) is sent on the same channel used to transport data traffic. Therefore, when a failure occurs along the data traffic path, both control and data traffic can be affected. This paper explains how failure recovery can be deployed in such a network. To achieve carrier-grade quality, the network should be able to recover from the failure within 50 ms. We apply two well-known recovery mechanisms -restoration and protection-for the control and the data traffic, and run extensive emulation experiments. The emulation results show that restoration does not allow to recover within 50 ms. Moreover, the restoration of the control traffic delays the restoration of the data traffic. The emulation results also show that protection for both control and data traffic can meet the carrier-grade recovery requirement, even in a large-scale network serving many flows.
