Bridging and Processing

[iomran]

I am an isp using 3640.I have frame relay subinterfaces.Some of them are in bridge groups.A bridge group is dedicated to only 1 interface.I noticed that when customers configured with bridge groups are active and getting much traffic,the processing of the router increases and this appears in the IP INPUT process.I am using cef and fast switching.For serial interfaces without BVI interfaces,a part of the output of the command "show interface stats" is almostly as follows :

Serial0/3

Switching path Pkts In Chars In

Processor 5830 11402

Route cache 36051 49785

Total 11881 31187

Which shows that the cache is used in forwarding many packets while for bvi interfaces,the output is as follows:

Switching path Pkts In Chars In Pkts Out Chars Out

Processor 19410142 399036124

Route cache 0 0

Total 19410142 399036124

Which shows that the cache is not used at all for BVIs.And I think this may be the reason of high processing.does any one have any suggestion to solve this problem.I think I can disable cef to force the interface to use fast switching.Does anyone has any suggestion

[tbaranski]

If BVI traffic can't be CEF'd for some reason, my understanding is that fast switching will be tried next automatically. Only if that fails as well will the packet be process switched.

Are you using a reasonably current IOS version? You may want to check with TAC to see what code versions (if any) can use CEF or fast switching for BVI traffic.

[iomran]

Thanks for your response.My IOS version is 12.2(17a).

I think my router is neither performing cef nor fast switching because as soon as the BVIs carry more traffic ,the processing increase.But what is amazing is that one active BVI increases cpu processing by about 20%.

Thanks alot