On Jul 9, 2012, at 12:14 AM, Ramana Reddy wrote: > Thanks for your reply. In the above setup, I believe that the other two OF > switches just actings as L2 learning switches, as they are not directly > connected to OF controller. If this is the case, how the OF switch ( the one > which > is connected to Controller) inserts the rules in the other two OF switches.
A switch doesn't insert rules in other switches. Only controllers do this. So I'm not sure what your thinking is here. If the two switches not connected to the controller are acting as learning switches, that's because they have some "fail open" behavior where they automatically do this if they can't contact their controller. > How in-band solves this problem. > Any help is this regard is more helpful to understand the internal operation > of in-band control. I don't think there are any simple answers here. For one thing, in-band control is not strictly defined -- it means different things to different people and to different switches. With your simple topology, the answer may be relatively simple. If the not-directly-connected switches expect to talk to the controller over a vlan that's configured on the port that connects them to the directly connected switch, you can probably just install a flow on the directly connected switch to send that vlan to the controller. Then the satellite switches should be able to reach the controller too. As a sidenote, this doesn't really seem to be NOX related. You may find the openflow-discuss mailing list or something more helpful. -- Murphy
