On 2007/02/25 20:05, Daniel Ouellet wrote:
I don't think you can do this exactly, but you can run OSPF on them,
let OpenBSD handle the main BGP sessions, and feed back a small BGP
table to the cisco containing just the prefixes that it needs to know
how to route. Something like this...
physical: peer -> cisco -> openbgp
e-bgp: peer <---------> openbgp (n.b. multihop for ebgp sessions)
ibgp: cisco <- openbgp
basically, cisco must know routes for any packets that will be fed
to it.
in some cases (e.g. one transit feed going into cisco) you may be
able to get away with just a static default route to the transit
on the cisco and OSPF or static routes back to your network.
Same with layer3 switches if you need more PPS than you can handle
on a PC and can live with limitations of the switches (e.g. restricted
table sizes and buffers).
I have ports for dynamips and dynagen if you need to play with cisco
configs and don't have spare ciscos: http://spacehopper.org/openbsd/
| Paul Jackson | Re: cpuset-remove-sched-domain-hooks-from-cpusets |
| James Bottomley | Re: Announce: Linux-next (Or Andrew's dream :-)) |
| David Miller | Slow DOWN, please!!! |
| Masami Hiramatsu | Re: [RFC PATCH v4] Unified trace buffer |
git: | |
| Gerrit Renker | [PATCH 27/37] dccp: Integration of dynamic feature activation - part 2 (server side) |
| Parag Warudkar | Re: 2.6.29-rc3: tg3 dead after resume |
| Jarek Poplawski | [PATCH] pkt_sched: Destroy gen estimators under rtnl_lock(). |
