RE: [nsp] [nsp] best practice for bgp load balancing with unequal bandwidth

From: Stephane Perez (perezs@nortelnetworks.com)
Date: Wed Mar 21 2001 - 05:22:13 EST


Hello,

  When you advertise routes out, you can use instead or with the AS Prepend
the MED which is specified out in the modifier. The lowest value of the MED
is the more preferred route.

I've also got a question about your design: Are the links are placed on the
same router or on different one ?

regards

----------------------------------------------------------------------------
----------------------------------------------------------------------------
----------------
PEREZ Stephane
Carrier IP Specialist
Nortel Networks
ATS IP Core Infrastructure
25, allee Pierre Ziller
06560 Valbonne, FRANCE
Phone: +33 4 92 96 18 05 Mail : mailto:perezs@nortelnetworks.com
Fax : +33 4 92 96 16 68 www : http://www.nortelnetworks.com
Esn : 296-1805 Intranet :
http://hector.europe.nortel.com

-----Original Message-----
From: Gert Doering [mailto:gert@greenie.muc.de]
Sent: 21 March 2001 10:56
To: Edward S. Desouza; Budi Widjojo; cisco-nsp@puck.nether.net
Subject: Re: [nsp] [nsp] best practice for bgp load balancing with
unequal bandwidth

Hi,

On Tue, Mar 20, 2001 at 10:03:25PM -0800, Edward S. Desouza wrote:
> You can influence incoming traffic in the scenario
> that you say using two techniques :
> 1. Advertise part of your pool on both links and
> advertise a less specific ( aggregate route ) to both.
> This will help in case your link to one ISP fails.

This shouldn't be done. Never advertise more specifics unless there's
a hard reason to - load balancing isn't. Look at the sorry state of the
global BGP table before giving advice like this.

> 2. Do an AS Path Prepend

3. Check with the upstream provider(s) what kind of "community" settings
   exist to control incoming traffic. This can often be finer grained
   than prepending.

gert

-- 
USENET is *not* the non-clickable part of WWW!
 
//www.muc.de/~gert/
Gert Doering - Munich, Germany
gert@greenie.muc.de
fax: +49-89-35655025
gert.doering@physik.tu-muenchen.de



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:32 EDT