peering charges?

Jun Wu (John) jun at wolfox.gsl.net
Mon Jan 27 20:00:58 UTC 1997


===== Vadim Antonov previously wrote: ====
> 
> 
> b) a 100-byte packet travelled from provider A to provider B.  Should A pay
>    to B or vice versa?
> 

Believe this is not the key problem with big players not be willing to
peer with small one. The key problem is WHERE they are exchanging traffic.
As long as both shares half of the load to haul traffic across the
continent/ocean, that is okay. But if the small provider has only one
local exchange, the big provider ends up hauling traffic both ways,
and the small one gets competetive edge in his local market because he
does not pay any wide area transmission cost. The cost would have been
paid if he bought a pipe from the big provider instead of free peering.
That is just a rough picture.

If the small provider, however small he is, buys a pipe across continent/ocean
and peer with big provider at multiple locations, there is really no
reason not to peer with him except some CPU/router-process concerns.

Jun





More information about the NANOG mailing list