routing policy questions

[masaki@merit.edu]

> Suppose, my pTLA is collection of 3 AS. 2 of them have tunnels.
> If I do aggregation, it will not only make paths much longer,
> I will be forced to export aggregate with different AS paths!!!

Hi. Alexey,

In your example, a pTLA is "shared" by more than one AS? In common
cases, I think prefixes are allocated "hierarchally"; a pTLA is
assigned to one AS (site) and the site delegates NLAs to the others.
I think that your example could be translated into that a site other
than root of pTLA wants to do multi-homing.

Masaki

>> From: kuznet@ms2.inr.ac.ru
>> Subject: Re: routing policy questions
>> Date: Tue, 8 Sep 1998 22:18:11 +0400 (MSK DST)
>> Message-ID: <199809081818.WAA18334@ms2.inr.ac.ru>
>>
> Hello!
> 
> > Recently, I established a tunnel with an European site that has a
> > /48 assigned from a pTLA (in US but not me) through some transit
> > site.
> > 
> > I'd like to ask people on 6bone who are involved in multi-homing
> > about how you implement it (keeping the 6bone routing policy).
> 
> It is pretty clear that this 6bone rule is in serious contradiction
> with bgp4 paradigm and with concept of AS numbers. Read, with reality 8)
> 
> Probably, I do not understand something, but it is impossible
> to inhibit exporting shorter prefixes without getting rid of
> these concepts.
> 
> Suppose, my pTLA is collection of 3 AS. 2 of them have tunnels.
> If I do aggregation, it will not only make paths much longer,
> I will be forced to export aggregate with different AS paths!!!
> 
> I am sorry, I prefer to violate any rules, only to avoid BGP breaking.
> 
> Alexey Kuznetsov
>