Dear Dr Peering,
Asia is a strange place to try and peer in. An area of many incumbent telcos who are quite happy to peer on the West Coast of the USA but not in Asia. They seem to be willing to fill the coffers of the global Tier1 providers and the TransPac submarine cable owners rather than improve intra-Asian latency and reduce cost for all in a desire to pretend to be a "regional" Tier1 and make everyone pay a toll to access their particular bit of the market. Yet, each one of them laments that others won't peer with them in Asia either and so their costs and latency are held high.
Are there same sage words you could offer to help break this hypocritical stalemate?
It does seem odd that a Tier 1 ISP in one Internet region would prefer to pay for the expense transoceanic circuit to the United States, and peer there openly, than to let someone else build into their own country and peer for free there. Or does it?
One reason was summed up quite well by John Milbourn (at DACOM at the time) :
“Protecting one’s home market is far more important than any benefit peering in one’s home market can provide.”
For the Tier 1 ISPs in a region, their home market is one where they pay no transit fees by definition. Anyone who wants to send traffic to their customers from anywhere in their home market will pay at least one of the Tier 1 ISPs directly or indirectly.
What would happen if a competitor (light blue circle in the diagram above) shows up in country and gets peering with a Tier 1 ISP in that region? The new ISP entering the market could cherry pick the best customers and offer transit prices below the prices the Tier 1 ISP has set. The entering ISP can simply send that traffic right over to the Tier 1 ISP’s network for free.
This exact situation happened when one International ISP went into Brazil, negotiated peering with one of the Tier 1 ISPs there, and was de-peered shortly after they started acquiring some of the Tier 1 ISP’s customers. Generally speaking, Tier 1 ISPs have a restrictive peering policy in their home Internet Region, and do not want any more peers. As Waqar Khan (Qwest) phrased the Tier 1 ISP perspective, “We have all the peers we need”, and he is correct -- a Tier 1 ISP already has access to all routes in the Internet Region.
Another example from the field. You will find that in peering negotiations, ISPs such as Korea Telecom will negotiate what Korea Telecom routes can and can not be announced where. In one case, peering was accepted only if the peer provided free transport from Korea to Hong Kong and peered with them there in Hong Kong, and under the terms that these routes would never be announced within Korea. So the peer could not compete in Korea. This example shows how protecting the home market trumps peering benefits and is negotiated accordingly.
The Tier 1 ISPs in Asia that have a restrictive peering policy in their home markets will often have an open peering policy on the west coast of the US. Why? The reason is that the peers on the west coast of the U.S. can’t compete with them in their home market; the latency is too high for the peer to be able to sell transit to the home market.
This is why you see a lot of Asian ISPs peering openly with other Asian ISPs in the U.S., and traffic between Asian countries following that boomerang path across the ocean twice.
Solutions? Many layers of political issues come into play that have prevented many initiatives except for bilateral peering across half circuits. Since most Asian countries wish to be the topological center of the Asian Internet, they are have a disincentive to helping one another be the topological core of the Asian Internet. DrPeering believes the solution is one that allows any Asian country to be a node in a distributed peering system that topologically replaces the U.S. West Coast peering.
DrPeering [at] DrPeering [dot] net