"Cleave Pokotea"
I am assuming there are two main methods of Inter-connection used in NZ; 1. Sats, 2. Southern Cross (?) cable. Is traffic classified (?) at the main entry points or are ISP's classifiying traffic via IP ranges (The dotted quad ranges seem's a rather ad-hoc method) and / or other methods?
Both techniques, i.e. counting traffic on separate circuits and counting packets based on source/destination addresses in common parts of the network are used by different ISPs and carriers. Sometimes both techniques are used for different services within the same ISP. Rating by IP address is usually done by monitoring routing tables; routes sourced from NZ peers are generally domestic, and can be charged as such. The actual charging is typically done by taking a feed from a router's flow cache (e.g. cisco's NetFlow), and then classifying the flow by customer and type based on the endpoints, and summarising the resulting data. The nice thing about that approach is that the data collection can be done anywhere in the network; on large networks, it's often best done at the edge (close to the customers), leaving a fairly dumb core. For traffic shaped connections, Barry Murphy's description is fairly typical; national traffic is fairly cheap, so ISPs tend to rely on the tail circuit to regulate traffic, and put traffic shapers on the international gateways, which select traffic by IP address. Otherwise, you're into having separate meterable paths to national and international sites. Drop me a line if you want more details on IP accounting techniques. -- don