Many ways to go about it. I'm not sure any is more correct than another. As long as you aren't wasting the router's CPU sending ospf hello's out all of the SVI interfaces.
For my personal preference, I try to enumerate the networks in the the ospf process. This allows me to quickly see all the networks being announced on that router just by sh run | b router ospf. The list may get long. but it is a definitive list of what that router is allowed to announce. You can always supernet if you have contiguous space, or break it off into another area, and range, if appropriate. I'm also a fan of using passive-interface default, and no passive-interface <interface> for the uplinks, in the routing process. Also for centralization. I can see exactly what interfaces should have OSPF neighbors. This also gives you the control to add additional SVIs, without any additional interface commands. I tend to treat redistribute connected as a bad practice, simply because, if an interface is created and IP'd, i don't necessarily ever want it to instantly start announcing its network through my routing protocol. One wrong subnet typo on a new SVI, can instantly cause issues... I'm sure you'll hear a few opinions on how to do it, as there are a few approaches. If there are cpu or network benefits for one way over another, I'd love to hear them. --Pete _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
