Hello all. I am running a simulation of SDN Switches using OpenDayLight as a controller. The network consists of 5 SDN Switches as seen in this topology. The destination node received the packets. The packets traveled through "route 1" as the controller algorithm detected the shortest path.
I wanted to introduce non-openFlow switches to test what would the controller do. So I added an Ethernet L2 switch "let's call it a dumb switch" because it doesn't require to have an IP or mac addresses. It behaves more like an extension cable. I added it between two SDN switches in "route 1" as seen in this image. The controller decided to avoid using this route this time and used the longer "route 2" for the packets to travel to their destination. It treated the Ethernet switch as a blockage maybe.
So I got curious, I added another "dumb" switch on "route 2" to force the controller to face either one of the dumb switches. The topology is seen here. So the packets did travel through the longer "route 2" eventually. But non went through the shorter "route 1".
Does anyone know how come the controller decided to use one dumb switch and ignore the other? Note that protocol packets travel normally through all SDN and dumb switches.
Software used:
Riverbed Modeler 18.8
OpenDayLight Oxygen 0.8.4
Ubuntu running on Virtual Machine
WireShark
No comments:
Post a Comment