Re: Resolving geo discussions

[Brian E Carpenter <brian@hursley.ibm.com>]

Yes, I think it is time to get out an info RFC on this. Indeed I haven't
seen any new arguments for a number of years. We need a brave person to
act as document editor though.

  rian

"Bound, Jim" wrote:
> 
> I have to throw in the towel and say I think Tony, Noel and Christian
> convinced me.  Geo as core method in our architecture is not going to
> fly nor should we try to push this as RFC std. But doing bottom up
> probably will happen is my guess.  I think an INFO RFC or BCP would get
> consensus here stating the ramifications of using this approach, but its
> your own trip.
> 
> I would suggest this would put this debate as lets document it and issue
> BCP and we move on to solve the problem at hand with architecture we can
> review.  Maybe if we had matrix of each one proposed on slide we could
> share as idea?
> 
> - We kill this discussion and turn it into work with deliverable.
> - We remove it to continue the architecture effort
> - And maybe we could produce a doc INFO/BCP that would be useful and the
> IETF would think we have some clue as a team to produce something others
> can use :--)
> - And we don't discuss it for 10 more years and just point to the RFC
> which will be useful to those who have children and come here to do work
> someday :--)
> 
> /jim
> 
> > -----Original Message-----
> > From: Christian Huitema [mailto:huitema@windows.microsoft.com]
> > Sent: Friday, April 11, 2003 4:24 PM
> > To: Iljitsch van Beijnum
> > Cc: multi6@ops.ietf.org
> > Subject: RE: Resolving geo discussions
> >
> >
> > > > We have had this discussion many times in the past 10 years. After
> > many
> > > > rounds of e-mail, the consensus generally is that if geo
> > is going to
> > > > happen, it will happen "bottom up" and be user driven.
> > >
> > > This is news to me. I would really appreciate some pointers to
> > archives
> > > of these past discussions.
> >
> > Try the Big Internet mailing archives:
> > ftp://munnari.oz.au/big-internet/list-archive
> >
> > > > For example, one
> > > > may see the chamber of commerce of lower Picardy request
> > a /32 from
> > > > RIPE, and use it to provides "independent" /48 addresses to its
> > > > members, using a combination of tunnels and arrangements
> > with local
> > > > ISP. In fact,
> > > > there are already a few virtual ISP operating under this model.
> > >
> > > So if I want to do this in a slightly larger area, say everything
> > > within a 20000 km radius of The Hague, I could get a larger address
> > > block, say a /16?
> >
> > Not really. You will be subject to the same slow-start
> > algorithm as regular ISP. You probably will get a /32 first,
> > and when you will have demonstrated that you serve more than
> > N sites, you will get a /30, etc.
> >
> > -- Christian Huitema