Creating exchanges

A few salient points to this discussion.

In terms of retail tariffs DISTANCE DOES NOT MATTER for International
Private leases. What matters is AGGREGATE VOLUME, which in turn
dictates retail pricing. Its cheapest to run leased lines to the point
where all the other lines run to. Distance is irrelevant. Accordingly
Guam and Hawaii are pretty shocking exchange locations - the fibre
facilities are, as pointed out, simply points where the photons come
up for air and a quick blow dry before taking another deep breath and
getting wet again.

For about the past 5 years the defacto Asia Pacific exchange location
has been the US west coast.


  - its the cheapest common termination point
  - the facilities there are ok
  - its the cheapest common termination point
  - the onward transit issue is far easier to solve
  - its the cheapest common termination point
  - the time penalty is about 120ms - which is ok
  - its the cheapest common termination point
  - its assumed to be politically stable

Now it would be REALLY GOOD to find an international exchange point
well within this region of the world rather than on the western
edge. In the longer term (and here I'm talking 5 - 10 years)
widespread use of use a facility could result in a cheaper and better
Internet facility for this region, and the greater dispersion of
traffic would remove critical earthquake-prone regions of the globe
from their role in underpinning a large hunk of Internet
connectivity. But to get from now to then is a process of interaction
which includes the undersea cable systems, Internet peering, transit
and settlement structures, technology-dictated degrees of freedom and
common strategic purpose to pull it off.

Personally, after working this issue for 3 1/2 years, international
neutral IXs will not work in Asia Pacific.

I don't share Barry's absolute pessimism here, but I think it will
require patience as well as effort.


   Geoff Huston