Re: The same prefix originated within two different ASs

Photon - NOL Staff <photon@dazed.nol.net> Tue, 11 July 1995 17:54 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa11274; 11 Jul 95 13:54 EDT
Received: from CNRI.Reston.VA.US by IETF.CNRI.Reston.VA.US id aa11270; 11 Jul 95 13:54 EDT
Received: from interlock.ans.net by CNRI.Reston.VA.US id aa12694; 11 Jul 95 13:54 EDT
Received: by interlock.ans.net id AA47851 (InterLock SMTP Gateway 3.0 for iwg-out@ans.net); Tue, 11 Jul 1995 13:37:15 -0400
Message-Id: <199507111737.AA47851@interlock.ans.net>
Received: by interlock.ans.net (Protected-side Proxy Mail Agent-3); Tue, 11 Jul 1995 13:37:15 -0400
Received: by interlock.ans.net (Protected-side Proxy Mail Agent-2); Tue, 11 Jul 1995 13:37:15 -0400
Received: by interlock.ans.net (Protected-side Proxy Mail Agent-1); Tue, 11 Jul 1995 13:37:15 -0400
Date: Tue, 11 Jul 1995 12:39:05 -0500 (CDT)
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Photon - NOL Staff <photon@dazed.nol.net>
X-Sender: photon@dazed
To: "Christopher D. Wheeler" <cwheeler@interglobe.com>
Cc: bgp@ans.net, bgpd@merit.edu
Subject: Re: The same prefix originated within two different ASs
In-Reply-To: <199507111613.AA47635@interlock.ans.net>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

On Tue, 11 Jul 1995, Christopher D. Wheeler wrote:

> 
> 
> Can someone please reiterate to me, from experience, reasons why a prefix
> shouldn't be advertised as originating from two separate ASs?  I think 
> this is a bad thing and I understand that by advertising the prefix from 
> two separate ASs there is no way to distinguish the policies of that 
> prefix.  But, in today's Internet, does anyone have any experience with what 
> kind of problems this equates to, so that I can forward them on to the 
> appropriate people.
> 
> I have a site that I am trying to convince that this is a bad thing and I
> would like some support.  The customer is a dual-homed site to provider A
> and provider B, but doesn't want to run BGP (fear of routing protocols I
> guess), rather just wants their prefixes advertised in both provider A and
> provider B ASs. 
> 
> Thanks.
> 
> -Chris
> 
> 
> -------------------------------------------------------------------------------
> Christopher D. Wheeler                 INTERNET: cwheeler@interglobe.com
> interGlobe Networks, Inc.	         OFFICE: 206.329.9266
> 1711 E. Olive Way, Suite 402		    FAX: 206.860.7821 
> Seattle, WA  98102                 	   INFO: info@interglobe.com 
> 				            URL: http://www.interglobe.com
> 
> 


If you just need a reason to convince your customer, then routing symmetry
is probably your best answer.  If providers A and B both announce the
customers network, when provider C needs to send a packet to the network in
question (Provider C is some distant network with no direct connections to A
or B), there is no clear way determine which path to take.  The benefits of
and desirablilty of routing symmetry have been outlined in numerous places
by numerous people.  

For that matter, how is this customer planning to do their own routing over
two links without using BGP or something similar?  Are they default routing
to provider A and doing a large amount of static routes to the customers
of provider B?  Really, the idea of them _not_ using an intelligent routing
protocol in this situation seems silly.

Brandon

.-------------------------------- ------ ----- ---- --- -- -  -   -   - 
|      _____    _______   __  
|     |___  \  |_____  \ |__|          Brandon Lee Black                 .
|      __ \  \  __   \  : __                                             .
| --==|  |=|  ||  |==|  ||  |==] TCP/IP Networking - Unix Admin [=====-- :
:     |  | |  |:  `\/'  :|  :---.                                        :
:     |__| |__| \______/ |______|           Photon                       | 
.          Networks On-Line                                              |
.                                        photon@nol.net                  |
       -   -  - -- --- ---- ----- ------ --------------------------------'