Hello All,

I have an x86 router (dual intel xeon 3.4) that is running about 150U / 350D of bandwidth through 2 providers. Each provider is a BGP peer and we are getting 2 full tables. Everything has been going very well for a very long time. We are starting to push more traffic and as a result the CPU is being used more. I was looking at the CPU usage with Profiler tonight and I noticed that there is anywhere from 10-15 percent CPU on queuing. We do not have a single queue on the system, only 1 input filter rule, no NAT, no connection tracking, no mangle, nothing of the sort really. I am wondering why CPU would be used on queuing then and if there are any other items that could affect this. I checked a similar setup core router (less the BGP) and the queuing on that box is less than 1 percent so I cannot understand why this box would have any more than that.

Any comments or tips would be appreciated!

Thanks all!

send a supout.rif file to support. either you missed something, or some process is mistakenly misattributed to queuing.

I have just sent three supouts of my routers - I see the same behaviour

by the way, what's 'routing'?.. is it bgp? I have ~0% on routing - we have only simple OSPF

Have anything came up?