design using duplicate ip addresses for redundancy


In designing a network with 2x 6500 cores, lets say about 10 4500 distribution (layer 2 only) and lots of edges.. I was testing with CEF. I find that if you move the root bridge away from the HSRP on the cores you are confined at layer two to send all packets over to your

6500 backbone link - and back again (first 6500 being root for most (but not all) vlans). This gives you a doubling of traffic across this link.

MLS on the other hand allows you to build flows and share those flows between your 6500's meaning whichever interface the next packets come into your core on they will be immediately switched. So it seems to me that CEF is slightly restricted (within the confines of layer 2) in its 'singlemindedness'.

For example if you swap the layer 2 bridge away from the layer 3 master all packets must travel to the layer three across the backbone.. the command show ip cef | inc [vlan gateway address] shows which path/router will be switching the packets..

Which brings me at last to ask.. what if I just duplicate the core IP addresses and MAC addresses on each interface and do away with HSRP?.. The distribution can block and forward layer 2 to whichever core is decided and now every packet will be immediately layer 3 switched on arrival at the core as opposed to having to cross the backbone link.

Tests in the lab have proven good.. now when you look at the show ip cef output you can see that they are both layer three switching...

Is there something wrong with this that I am missing because it seems optimal to me?!



Reply to
Loading thread data ... Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.