[c-nsp] Redundant default route, without round-robin routing

schilling schilling2006 at gmail.com
Fri Sep 28 10:59:05 EDT 2007


Actually this scenario is discussed in detail in Internet Routing
Architecture by Sam Halabi.

If  customer border routers are not directly connected, routing loop may
occur. the customer BGP routers should advertise a default to internal OSP
cloud only if it gets a default from it's eBGP peer implemented by a
route-map.

If customer border routers are directly connected,  things are  very hard to
go wrong.

schilling

On 9/27/07, Dario < dario.donsion at soporte.rediris.es> wrote:
>
> Dear all,
>
> I've configured something similar a few days ago with one customer running
> OSPF
> as internal protocol.
>
> We've installed two links between us and the customer (two different
> border routers),
> one acting as backup (only active when the main one fails).
>
> We've configured two BGP sessions with each of the customer routers, using
> local
> preference to prefer the main link. Also the customer uses local
> preference to prefer
> the main link. We only advertise default route and we receive all the
> customer
> routes (the customer forces the advertisement using static routes to null
> with a
> poor metric).
>
> Inside the customer it's configured an iBGP session between the two border
> routers
> in order to pass the default route between them. Then, each border router
> receives
> the default route via eBGP (from us) an iBGP (from the other border
> router). With
> the local preference, the main link is prefered.
>
> The defaul route inside the customer is distributed in the OSPF cloud
> using
> "default-information originate allways".
>
> At normal situation the main link is used: all traffic inside the OSPF
> cloud reach
> one of the border routers, and then the border router send it directly to
> us or (if it's
> the backup one) to the other border router.
>
> If the main link fails all customer traffic flow to the backup border
> router and then
> to us.
>
> The solution works fine ;)!
>
> Regards,
>
>         Dario D.
>
>
> El Jueves, 27 de Septiembre de 2007 17:05, Jon Lewis escribió:
> > On Thu, 27 Sep 2007, Vincent De Keyzer wrote:
> >
> > > I was thinking of having both BGP routers advertising a default route
> in
> > > OSPF; but in this case, in order to limit packet out-of-sequence
> problems, I
> > > want to make sure that every IP flow uses the same default gateway.
> >
> > Have you tried having one of the eBGP routers originate a default route
> > into OSPF using a different metric than the other?
> >
> > #default-information originate ?
> >    always       Always advertise default route
> >    metric       OSPF default metric
> >    metric-type  OSPF metric type for default routes
> >    route-map    Route-map reference
> >
> > ----------------------------------------------------------------------
> >   Jon Lewis                   |  I route
> >   Senior Network Engineer     |  therefore you are
> >   Atlantic Net                |
> > _________ http://www.lewis.org/~jlewis/pgp
> <http://www.lewis.org/%7Ejlewis/pgp> for PGP public key_________
> > _______________________________________________
> > cisco-nsp mailing list  cisco-nsp at puck.nether.net
> > https://puck.nether.net/mailman/listinfo/cisco-nsp
> > archive at http://puck.nether.net/pipermail/cisco-nsp/
> >
> >
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>


More information about the cisco-nsp mailing list