From: Stanislav Sinyagin (SSinyagin@xxxxxx)
Date: Mon Dec 06 1999 - 06:56:23 GMT-3
First of all, you should carefully read Sam Halabi's article on Cisco CD:
Technology Information ->
Internetworking Case Studies ->
Using the Border Gateway Protocol for Interdomain Routing
Then, you would keep in mind that IBGP keeps next-hop attribute unchanged, unle
ss
next-hop-self is specified. Thus, it does not matter wether we point the
IBGP neighbor to other's loopback or to a physical interface --
the BGP updates will come with next-hop attribute pointing to the external
router's physical interface.
In EBGP, it is important that the routes coming into my router from a
foreign one have next-hop to an address which I am aware of.
Imagine: you're administering AS701 (UUNET), and your friend Stan
administrates AS8359 (MTU-Inform). The physical link has addresses
1.1.1.1 (UUNET) and 1.1.1.2 (MTU). So, if I (Stan) receive updates from UUNET
with next-hop set to 1.1.1.1, then it's all right, I know how to reach
1.1.1.1, 'cause it's my connected route.
Otherwize, say UUNET's updates come with next-hop set to 2.2.2.2. I have no ide
a
of 2.2.2.2, and thus I can't establish these routes into my routing table.
And I should say "ebgp-multihop", and establish a route to 2.2.2.2 in some way.
I hope my explanation isn't too deem. Halabi explains it in much more clear way
.
Stan
----- Original Message -----
From: Peter Van Oene <vantech@sympatico.ca>
To: <ccielab@groupstudy.com>
Sent: Sunday, December 05, 1999 08:44
Subject: Re: Re[2]: BGP Update-source
> I'm certainly no expert in BGP, however whenever I use looback addresses
> (which is whenever I use BGP) I always use the EBGP-Multi-hop statement.
> Given that the two loopbacks are essentially a minimum of 2 hops away, I
> would see that this command is relevant. How would the router differentiate
> it from any other network that was not directly connected?
>
>
>
> Peter Van Oene
> Senior Systems Engineer
> UNIS LUMIN Inc.
> www.unislumin.com
> Convergis Member Company
> www.convergis.com
>
> ----- Original Message -----
> From: Stanislav Sinyagin <SSinyagin@mtu.ru>
> To: Martin Bander <cisco103@hotmail.com>
> Cc: <honsiong@hotmail.com>; <ccielab@groupstudy.com>
> Sent: Sunday, December 05, 1999 11:14 AM
> Subject: Re[2]: BGP Update-source
>
>
> > Ebgp-multihop is not required at all in this scenario. Your bgp
> > session is "Active" because one of your routers does not know how to
> > reach the other's loopback. You should tell it by static or dynamic
> > routing. And make sure that both point to each other's loopback and
> > have update-src loopback, or both point to other's physical interface
> > (and no updare-src at all).
> >
> > Regards,
> > Stan
> >
> > Martin Bander <cisco103@hotmail.com> wrote:
> > MB> Don't forget the ebgp-multihop comand, since r1's loopback interface
> is not
> > MB> 'directly connected' to r2.
> >
> >
> > MB> ----Original Message Follows----
> > MB> From: "hon-siong chan" <honsiong@hotmail.com>
> > MB> Reply-To: "hon-siong chan" <honsiong@hotmail.com>
> > MB> To: ccielab@groupstudy.com
> > MB> Subject: BGP Update-source
> > MB> Date: Thu, 02 Dec 1999 20:28:02 PST
> >
> > MB> I followed Sam Halabi book on configuring a simple BGP peering by
> pointing
> > MB> to a loopback interface. Scenario is as simple as follows:
> >
> > MB> -----R1 --------R2
> >
> > MB> Where R1 has a loopback interface and both are in same AS. In R1,
> command
> > MB> as:
> >
> > MB> neighbor <R2> remote-as 100
> > MB> neighbor <R2> update-source loopback 0
> >
> > MB> The peering can never be established since then. A "Sh ip bgp neigh"
> showed
> > MB> "Active" status only?!
> >
> > MB> What's wrong?
> >
> > MB> Thanks in advance....
> >
> > MB> HonSiong
> >
> >
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:21:58 GMT-3