Fw: [ppml] 2005-1:Business Need for PI Assignments

[Michael.Dillon at radianz.com]

> For a concrete example, say each state is given a geographical registry

Aligning addressing to state boundaries is *NOT* geographical
addressing, it is geopolitical addressing. That is what the ITU
has proposed and I believe that it is the wrong way to go.

>A provider in Detroit may reasonably choose to get its connectivity 
solely
> from Chicago and Cleveland.  Either (a) the provider uses Illinois or 
Ohio
> addresses, meaning their customers can only multihome (or change 
providers)
> to other providers also serving Illinois and Ohio, or (b) the provider 
uses
> Michigan addresses and must announce more-specifics into the global (or 
at
> least US) routing tables for each customer block.  Both scenarios are 
worse
> than what we have today.

First, geographical addressing in IPv6 should be implemented
as an alternative choice. That way, there will be a competitive
marketplace of sorts, in which providers and companies can
choose whether or not to use geographical addressing. If the
geographical boundaries were as you described, then you are
simply describing a scenario in which a provider, of their
own free will, makes a dum choice. Market forces will sort
it out.

However, if Chicago is considered to be a regional center for
the purposes of geographic addressing, it is highly likely
that Detroit would be considered to be part of the Chicago
region. Geographical addressing ignores national and state
boundaries and looks at the cities which are centers of
commerce and transportation and communications. It is the
relationships between these cities(nodes) that will determine
the regional and sub regional boundaries. In the USA, the work
done on LATAs will be useful to guide this. However, anyone
looking for a system that carves up the world into sharply
delineated regions will be disappointed. Since geographical
addressing deals with the regions created by human commerce 
and communications links, the majority of "boundaries" will
be as fuzzy as those links. 

> The only way to make this work is to force providers (by law) to peer 
with
> or purchase transit from all other providers in a given location to be
> allowed to offer service there.

I can't cure insanity and I can't understand why people
think that technical problems need a top-down decree from
the "powers that be".

If and when IPv6 geographical addressing comes into being
it will be based on sound research and the consensus of
domain experts including the IP networking community and
the economics community and the anthropology community and
the geographic community.

> Unfortunately, many others proposing the same addressing model are 
planning
> to use it to replace, not augment, the current model.

It makes no difference to me what other people are
"planning" to do. Many people in the world plan impossible
things and most of them fade into the sunset.

> > Topology does follow geography.
> 
> No, it does not.  The vast, vast majority of networks I personally have
> worked on have little if any correlation to geography at the IP level.

You are just viewing the network in the wrong way. You
are looking at the topology of ASes. I am looking at the
topology of traffic flows and it makes little difference 
to me whether the flows between Boston and New York are
inside an AS or between ASes. 

Current IPv6 addressing mimics IPv4 addressing and aggregates
addresses by AS, more or less. I want to cut the pie in a
different way so that addresses can be aggregated by geography.
And just like Pizza Hut's square slice pizzas, I want to give
people a choice. Some people will choose the square slice and
some will go for the traditional wedge. Today, in IPv6 addressing
this choice does not exist.

> Geographic addressing makes things simpler for sites that are connected 
only
> to other sites in the same area, and that may be nice for consumers and 
some
> businesses, but it makes things worse for operators of more complicated
> networks.

No, it makes things *BETTER* for the operators of complicated
networks because they will not need to deal with the large number
of routes that are hidden behind the geographical aggregates.
Geographical addressing is an *OPTION* which providers can use
if it makes sense for them. If a provider does not use geographic
addresses in their network and they cover all of the USA, then
they can simply accept a single global aggregate route from their
peers. If the provider's network spans continents, then it makes
more sense to accept a half dozen continental aggregates from
peers in the relevant continents. That is all that changes for
a network that does not use geographical addresses internally.

--Michael Dillon