Cisco Support Community
Community Member

[unicast flooding] arp-timeout vs. mac-aging-time

hi, all:

i ran into a problem last night. it has to do with asymmetric routing and the above mentioned timeouts.

it took a while to figure out the fact that i was running into a unicast flooding problem. thank god for sniffers. anyway, i came across the following document:

i finally followed this document's recommendation and configured 'similar' timeouts for both arp and the macaddress table. things have settled down now, but i was wondering if anybody out there has seen this problem and if you would recommend actually giving the mac-aging-time more time (say 5 hours) so that arp works its magic first and unicast flooding is mitigated.

any comments will be appreciated.




Re: [unicast flooding] arp-timeout vs. mac-aging-time

I feel having a larger value for mac aging time is fine to resolve this problem, and don't see any major issues. But, you may have to think of any negative effects that might occur like using more memory to retain the MAC addresses and packets being forwarded to a machine even if it is not available etc. Ofcourse, you might be able to mitigate certain problems by increasing the aging time only for certain vlans. I am also wondering if VTP pruning will also help in anyway if a machine is no more available on the network. But, the solution suggested in the document gives best of both in my opinion, unless you are concerned about flooding frequent ARP broadcasts.

CreatePlease to create content