[ncc-regional-middle-east] Regional Peering
John Leong leong at qatar.cmu.edu
Mon May 22 22:22:40 CEST 2006
One last note on NAP and peering in general ... While my first message suggest that it is strange from an engineering point of view to have traffic between neighbouring country to travel all the way through US, there is the business side of the equation. Using an extreme example to illustrate a point, if 99% of my traffic is to the US and only 1% to my neighbouring ISP, and if it costs me serious money to establish a direct link to my neighbouring ISP so I can do the peering (plus whatever money I have to pay in a bi-lateral settlement), then I may say too bad to that 1% and hey can take the long route since it is not worth it for me as a business. Of course the ideal case is if there is a NAP in the neighbourhood that is already hosting the big US and/or European ISP. In which case, I would peer with my US partner there instead of paying for a long haul link all the way to the US. If everyone think that way, that NAP will attract everyone there ... including my neighbour. In that case, the cost to solve my 1% traffic to my neighobour problem can be pretty cheap since I no longer need to pay for an addition link to that neighbour (only whatever bi-lateral settlement we can work out which could be none if we are truely 'peers'). Best regards, John Leong.