[29996] in North American Network Operators' Group
RE: Running BGP4 on a Core Router
daemon@ATHENA.MIT.EDU (Dmitri Krioukov)
Wed Jul 12 14:13:39 2000
From: "Dmitri Krioukov" <dima@krioukov.net>
To: "Chris Liljenstolpe" <chris@cw.net>
Cc: "nanog" <nanog@merit.edu>
Date: Wed, 12 Jul 2000 14:25:48 -0400
Message-ID: <NCBBIKACLKNMKDHKKKNFMEJCEMAA.dima@krioukov.net>
MIME-Version: 1.0
Content-Type: text/plain;
charset="US-ASCII"
Content-Transfer-Encoding: 7bit
In-Reply-To: <110295.3172374783@res1-dialup60.ie.cw.net>
Errors-To: owner-nanog-outgoing@merit.edu
that is, your ip core and edge are sort of
very "close" to each other. it's not like,
say, in uunet, where this architectural
distance is greater. true?
--
dima.
> -----Original Message-----
> From: Chris Liljenstolpe [mailto:chris@cw.net]
> Sent: Wednesday, July 12, 2000 7:13 AM
> To: Dmitri Krioukov; Bora Akyol
> Cc: nanog
> Subject: RE: Running BGP4 on a Core Router
>
>
> Dimirtri,
>
> I have to disagree with you on this statement. While the
> physical trunks
> are not connecting routers together in the C&W network, there is most
> definately an IP core. The L2 core provides a mesh for core routers at
> each site, wich provide the hierarchy for edge/fannout routers.
> Therefore
> we have both an IP and L2 core with the IP core overlayed over
> the L2 core.
>
> Chris
>
>
> --On Tuesday, 11 July 2000 21.54 -0400 Dmitri Krioukov
> <dima@krioukov.net>
> wrote:
>
> >
> > we can even imagine some core that is
> > not ip core but, say, atm core and all
> > lsrs are atm-lsrs.
> >
> > actually some providers (like c&w)
> > have exactly this no ip core, overlay
> > model. it's far from being the best one.
> > --
> > dima.
> >
> >> -----Original Message-----
> >> From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu]On Behalf Of
> >> Bora Akyol
> >> Sent: Monday, July 10, 2000 12:10 AM
> >> To: nanog
> >> Subject: Re: Running BGP4 on a Core Router
> >>
> >>
> >>
> >> Even with MPLS, you need to run some sort of a routing protocol.
> >>
> >> ISIS or OSPF with TE extensions would do.
> >>
> >> One can also use BGP with MPLS Label extensions as well. By
> the way, how
> >> does this work with route reflectors?
> >>
> >>
> >> Bora
> >>
> >>
> >> ----- Original Message -----
> >> From: "Jesper Skriver" <jesper@skriver.dk>
> >> To: "HANSEN CHAN" <hansen.chan@alcatel.com>
> >> Cc: <nanog@merit.edu>
> >> Sent: Sunday, July 09, 2000 8:20 PM
> >> Subject: Re: Running BGP4 on a Core Router
> >>
> >>
> >> >
> >> > On Sun, Jul 09, 2000 at 07:49:37PM -0400, HANSEN CHAN wrote:
> >> > >
> >> > > Hi folks,
> >> > >
> >> > > I was hearing that typically BGP4 is run on all routers
> inside a POP,
> >> > > including access routers connecting to customers, border routers
> >> > > connecting other ISPs and core routers connecting to other POPs in
> >> > > the same network.
> >> > >
> >> > > I can understand why BGP4 is run on access and border routers. But
> >> > > running BGP4 on core routers is beyond my understanding. I thought
> >> > > you don't need to run BGP4 on core routers which are
> considered to be
> >> > > interior nodes.
> >> > >
> >> > > Can someone shed some light on what is the benefit of
> running BGP4 on
> >> > > the core routers?
> >> >
> >> > If these routers run "normal" ip routing you have to, as each router
> >> > does a lookup of the destination ip address of each packet,
> and forward
> >> > it accordingly.
> >> >
> >> > If you run MPLS, you don't have to, as it uses labels to get to the
> >> > next-hop router.
> >> >
> >> > /Jesper
> >> >
> >> > --
> >> > Jesper Skriver, jesper(at)skriver(dot)dk - CCIE #5456
> >> > Work: Network manager @ AS3292 (Tele Danmark DataNetworks)
> >> > Private: Geek @ AS2109 (A much smaller network ;-)
> >> >
> >> > One Unix to rule them all, One Resolver to find them,
> >> > One IP to bring them all and in the zone to bind them.
> >> >
> >>
> >
> >
>
>
>
