Re: Few comments on draft-ietf-multi6-architecture-02

[Thierry Ernst <ernst@sfc.wide.ad.jp>]

Hi Marcelo,

> i just want to make a comment about one of the points that you are 
> considering (this may be OT for this list though... but i hope that the 
> chairs will let me know when to shut up :-)
> 
> El 02/11/2004, a las 11:15, Thierry Ernst escribió:
> >   [BTW, just being curious, can't we consider a NEMO multihomed to 2
> >   distincts ISPs as a site ? cf 
> > draft-ietf-nemo-multihoming-issues-01.txt]
> >
> 
> I am not sure that all the multihomed nemos can easily fall under the 
> definition of a multihomed site.

Sure, some case do fall under the definition of a multihomed site, some
don't.

> IMHO, there are cases where it is clearly so, and that the multi6 
> solution will provide a solution for the multihomed nemo, while in some 
> other cases, i am not so sure.
> 
> For instance, suppose a nemo that has a multihomed home network. So the 
> nemo has two prefixes, one per ISP of the home network. In this case i 
> guess that it is clear that the multi6 solution should apply to this 
> network, i guess.
> 
> The same case could be when the nemo has two home networks (which i 
> guess that is the case that you are considering)
> 
> I mean in all these cases, the point is that the multihoming doesn't 
> occur between the nemo and the fixed network, but between the 
> attachment of the home network(s) and the internet. I guess that in 
> those cases, the multi6 solution should apply.

OK for the 1st case, but in the second case (the nemo has 2 home
networks), I would rather see the multihoming occuring between the
nemo and the home network. 
      
         home network 1
             |  _AR
  |  _-p1/56-|-|_|- p1/48 -- Internet
  |-|_| MR      _
  |   -p2/56-|-|_|- p2/48 -- Internet
             |   AR
         home network 2
  
    
> However, when we are considering a nemo that has multiple attachments 
> points, and for instance only one nemo prefix, then the fundamental 
> assumption of multi6 solution is no longer valid, since the nemo has a 
> single prefix. I guess that in these cases, the usefulness of the 
> multi6 solution will diminish.

Agree. In such case, I wouldn't consider the nemo as a "site" therefore
the site-multihoming solution wouldn't apply.

> But anyway, i fell that it is very important to determine which parts 
> of the nemo multihoming problem can be addressed with multi6 and which 
> parts will require additional tools.

Identifying which parts can be addressed by existing solutions of WG is
something that we should definitely clarify in
draft-ietf-nemo-multihoming-issues.

Thierry.