[ppml] YAGAvPBAFF (was: 2005-1:Business Need for PI Assignments)

David Conrad david.conrad at nominum.com
Thu Apr 28 17:43:58 EDT 2005


[Yet Another Geographic Addressing vs. Provider Based Addressing Food 
Fight, in case you were wondering]

Michael,

On Apr 28, 2005, at 5:30 AM, Michael.Dillon at radianz.com wrote:
> Since a geographic addressing model does provide for planned 
> aggregation
> at city and regional and continental levels, it can accomodate 
> exponential
> growth in multihoming when the multihomed organization buys services
> from two or more ISPs in their city who support geographical 
> addressing.

As I understand it, one of the traditional difficulties with this model 
is that it would require, in order to make business sense, the 
equivalent of PSTN-style settlements to deal with transit issues.  
While this approach would definitely be favored by the ITU and many 
PTTs, I'm not sure anyone else is particularly interested in recreating 
the settlements model.  Is this not an issue?

As an aside, I'm probably too cynical, but I have a sneaking suspicion 
this is one of the reasons the ITU is pushing national based 
allocations.

> Topology does follow geography.

No it doesn't.  Physical topology does (sort of by definition), but 
telecommunications network topology follows economics and politics 
which sometimes follow geography.

Until fairly recently, the topology of the Asia Pacific region was a 
star with the US in the center because in general it was cheaper to go 
through the US to send intra-AP region traffic than it was to link AP 
region countries directly.  I remember when two universities in 
Thailand which were no more than a few miles from each other exchanged 
traffic via Falls Church, VA.  The situation is better now, but there 
are still cases where folks in, e.g., one African country prefer to 
transit Europe to reach another African country.

If bandwidth on fiber going from NY to LA via Kodiak Island becomes 
cheaper than fiber going from NY to LA via Kansas City, I don't believe 
the network topology will follow geography.

Rgds,
-drc




More information about the ARIN-PPML mailing list