[6bone] RFC2772 rewrite -- bigger scope goals

[Bjørn Mork]

Nicolas DEFFAYET <nicolas.deffayet@ndsoftware.net> writes:
> On Mon, 2002-11-18 at 17:45, Bjørn Mork wrote:
>> Pekka Savola <pekkas@netcore.fi> writes:
>> > On Mon, 18 Nov 2002, Bjørn Mork wrote:
>> >> Nicolas DEFFAYET <nicolas.deffayet@ndsoftware.net> writes:
>> >> 
>> >> > BGP choose the route(s) with shorter ASpath and lower MED.
>> >> 
>> >> MED won't matter unless you have multiple links to the _same_ AS.
>> >
>> > unstated background is that Mr Deffayet is using "always-compare-med".
>> 
>> With 90+ uncoordinated peers? Doesn't that result in more or less
>> random routing? 
>
> http://noc.ndsoftwarenet.com/stats/aspath-tree/bgp-page-complete.php
>
> Our routing is very clean and not random.

Sorry, the word I was looking for wasn't really "random" but
"arbitrary".

You prefer routes with MED to routes without MED (or vice versa if you
are using Cisco defaults). I don't think most of your peers would
expect them to suddenly become the preferred path just because they set
a default MED.

What advantages do route selection based on MED give you?


Bjørn