International Internet Connectivity

steveb at steveb at
Tue Jan 6 11:07:49 UTC 1998

     I can give you the historical situation, although this is slowly 
     changing as the Net is becoming slightly less 'US-centric'.
     The philosophy for international connection to 'the Internet' (ie USA) 
     was exactly the same as with in the US itself.  That is, those wishing 
     to connect could do so at some financial cost levied by the US ISP/NSP 
     in addition to the cost of the transport.
     In the international case, this means a huge cost to those wishing to 
     connect as they must pay the whole cost of trans-oceanic capacity.  In 
     general, because of the fact that the Internet is still US-centric, 
     this results in high traffic flows outgoing from the US compared with 
     much smaller incoming traffic flows.  In the the case of New Zealand, 
     only aobut 20% of the total traffic is incoming at the US end.
     Traditionally (and I speak for Oceania only), there were virtually no 
     peering agreements internationally, with the result that the cost of 
     all traffic and all transport was borne by the country connecting to 
     the USA.  This is changing slowly as large US ISP's find the need to 
     service their multi-national customers in countries other than the US.
     To answer your questions more specifically:
     1) I'm not sure of the answer you want here, but in general the 
     peering rules internationally are similar to those in the US; that is, 
     the international connection would appear in a similar manner to any 
     normal ISP connection.
     2) Peering agreements are now being considered internationally and 
     these are generally very much 'ad hoc'.  There are no hard and fast 
     rules and the agreements are generally based on the perceived benefit 
     to each of the parties.  This may be done purely on a 'guesstimate' of 
     traffic ensuing from requests emanating from each of the parties and a 
     settlement made on this basis.  The peering arrangement may be at 
     either end of the international links although the 'normal' situation 
     is for the agreement to be held at the 'busy' end - normally the USA, 
     in the case of that country.
     3) The international connectivity (cables and satellites) is owned by 
     various consortia and capacity is 'sold' to interested parties.  This 
     may be done on a 'wholly-owned', end to end basis (depending on 
     connection and operation agreements in place between the countries) or 
     on a 'half-circuit' basis where the purchase of the (say) cable would 
     be made by one party. For example,  New Zealand would purchase the 
     capacity to half-way across the Pacific, in the case of a New 
     Zealand-USA connection.  The remainder, from that point in the Pacific 
     to the US, say, would be purchased by some US ISP/NSP.  Traditionally 
     the American half would be leased back to the remote country, who 
     would then effectively pay for the whole transport cost.
     Hope this helps...
     Steve Barr
     Netway Communications
     +64 9 357-1566

______________________________ Reply Separator _________________________________
Subject: International Internet Connectivity
Author:  George McCormick <gmccormick at> at internet
Date:    06/01/98 09:09

I would be extremely appreciative if someone could explain to me how US 
ISPs and backbone operators address their international connectivity 
o How is their traffic routed?
o How do peering relationships work and where do they take place? 
o Who pays whom for international transport(undersea cable and/or 
satellite) and services?
George McCormick
Cambridge, Massachusetts

More information about the NANOG mailing list