From: John Conzone (jkconzone@xxxxxxxx)
Date: Fri Apr 14 2000 - 18:58:11 GMT-3
It means find the destination. This is implemented to prevent non-meshed
IBGP networks from dropping routes bewteen AS's.
I'll try to give an example of what its designed to prevent.
You have to IGP session bewteen two routers, but they are not directly
connected, and the IBGP session is over an IGP, like OSPF. The routers in
between the 2 IBGP peers are not running BGP.
So one IBGP router talks to a remote AS, and the other IBGP router also
talks to a seperate remote AS. Now, if one of those remote AS's sends an
advertisement to one of the IBGP peers, it will pass it to the other IBGP
peer, who will advertise it to his AS. So far so good.
Now the remote AS that got the advertisement sends traffic towards that
advertised network to the BGP peer he learned it from, who needs to pass it
to his IBGP peer to get to its destination. But, to get to his IBGP peer, he
has to pass across a non-BGP router, who doesn't have a clue as to the
destination of the packet, since it is not a BGP router. Hence, it will drop
the packet, UNLESS it is in the IGP.
That's what BGP synchronization is designed to prevent. Sorry if this is
confusing. Someone please jump in if you can explain it better.
Cliff, I would also suggest Internet Routing Architectures from Cisco
Press. Also known as the "BGP Bible". Its by Halabi, and it is the BEST
reference for BGP I have ever read. This will all come clear to you if you
read that book!
----- Original Message -----
From: Cliff Webster <cliff@dcjob.com>
To: 'John Conzone' <jkconzone@home.com>
Sent: Friday, April 14, 2000 5:31 PM
Subject: RE: EBGP and IBGP
> I was under the impression Synchronization meant "Find the next hop in the
> IGP" not "find a duplicate route in the IGP." So with this rule you would
> require mutual redistribution or one way?
>
> Thank you,
> Cliff Webster
> Webster Data Communication, Inc.
> Pager (301) 303-0662
>
> -----Original Message-----
> From: John Conzone [SMTP:jkconzone@home.com]
> Sent: Friday, April 14, 2000 4:37 PM
> To: Cliff Webster; ccielab@groupstudy.com
> Subject: Re: EBGP and IBGP
>
> Cliff, when you talk about BGP synchronization, it means that
> BGP will
> check your IGP for routes to other AS's. If it doesn't find it, it
> drops the
> advertisement.
> No sync turns off the "find it in the IGP" algorithm, and allows
> BGP to
> advertise routes to other AS's not known to the IGP.
> This is normal operation of BGP.
>
> ----- Original Message -----
> From: Cliff Webster <cliff@dcjob.com>
> To: <ccielab@groupstudy.com>
> Sent: Friday, April 14, 2000 2:09 PM
> Subject: EBGP and IBGP
>
>
> > In the Lab I am doing BGP a scenario with 6 routers in a straight
> line. I
> > take 2 of the middle routers and do IBGP. The outside routers are
> doing
> > EBGP only. I have been under the impression that if I am using an
> IGP
> like
> > OSPF between the two internal routers, the protocols will
> synchronize on
> > their own (OSPF and IBGP), and EBGP routes will be passed. When I
> do
> this,
> > my routes will not propagate past the IBGP section in either
> direction.
> The
> > end EBGP systems are not exchanging routes. If I go to the two
> IGBP
> routers
> > and type NO SYNC. Everything works fine. Shouldn't OSPF and IBGP
> sync up
> > on there own- without the NO SYNC command?
> >
> > I am studying from the Bruce Caslow book. It is not real clear on
> the
> > subject.
> >
> > Thank you,
> > Cliff Webster
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:23:13 GMT-3