Novel fast switchover on OpenFlow switch

The specifications of OpenFlow switch define the fast failover¹ group table, which allows the switch quickly adapt to failure based on its local states. In a failure scenario, the traffic on a faulty path is fast switched to a predetermined path in a so-called active/standby manner. That is likely the standard and only way of achieving fast switchover. In this paper, we introduce a novel method for fast switchover/failover on the OpenFlow switch. The conceptual idea is that the new switchover follows an active/active manner rather than the active/standby one in the standard. Specifically, instead of using the fast failover group table we leverage the another one named select. By doing so, the traffic of an end-to-end connection is striped over multiple active paths on a weighted round robin (WRR) fashion. Whenever a failure happens across the paths, the data flow on a failed path is rapidly switched over to the remaining active paths. We modify an OpenFlow 1.3 switch to include the proposed switchover and evaluate the new switch under failure scenarios. The new OpenFlow switch has a better resource allocation as well as less traffic lost during the recovery process than the standard one.