North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Re: AGIS/DIGEX

  • From: Paul A Vixie
  • Date: Sun Nov 03 18:45:29 1996

[Somebody]
> [...]  Since Digex is at too few NAPs to peer with AGIS, and they
> are not at the CIX, we see no routes from them. 

[Paul Vixie]
> CIX is taking new connections, either T1, PB-SMDS, or DEC-PAIX (FDDI/Ether).

[Enke Chen]
> I have been thinking about a couple of issues regarding the 
> CIX Inter-exchange, and your comments would be appreciated:
> 
> (1) CIX is an association of ISPs. The CIX Inter-exchange puts 
>     CIX in direct competition with its members (ISPs) for 
>     customers.  Is there a contradiction here?   
> 
> (2) The CIX Inter-exchange has been used by some to bypass 
>     peering policies. I am wondering if the CIX Inter-exchange
>     remains a viable option in the peering policy review.
> 
> (3) Today there are quite a few inter-exchanges in the Bay Area
>     that are superior (in terms of tech. and richness in connectivity)
>     than the CIX Interexchange.  Is it fair to say that the 
>     CIX Inter-exchange has fulfilled its historical mission and 
>     is no longer needed? 

Enke,

(1) CIX is a membership organization which facilitates voluntary
connectivity between its members.  CIX doesn't have a product or a service,
it's a nonprofit industry trade association more or less on legal par with
the California Egg Association.  The contradiction is in your description
not the facts.

(2) If MCI doesn't like the routes CIX sends them, they should not accept
those routes.  If MCI doesn't want CIX to hear certain routes, they should
not send them.  If MCI derives no benefit from CIX, they should consider
dropping either their CIX connection or their entire membership.  (I hope
that they don't -- the industry as a whole, which CIX's charter says it is
here to do good things for, benefits from the kind of "peering policy bypass"
that I think you're complaining about here.)

(3) The other peering points are homogeneous.  You need to buy a circuit from
Pac Bell or MFS in order to join either of those peering points, and there is
no level-3 connectivity.  A lot of ISP's, especially international ones, find
one or both of those requirements onerous.  I do see a continuing mission for
CIX, else I would not still be consulting for the association.

In general, there is a mistaken view that L1/L2 exchanges can be cooperative
but that L3 exchanges have to be competitive.  Neither definition is accurate.

Thanks for bringing these important issues to everyone's attention, I'm always
glad to have a red carpet rolled out in front of my views on this subject.

Paul
- - - - - - - - - - - - - - - - -