[2679] in Commercialization & Privatization of the Internet
Re: Settlements
daemon@ATHENA.MIT.EDU (John Curran)
Thu Apr 9 21:43:54 1992
To: tom_limoncelli@warren.mentorg.com
Cc: com-priv@psi.com
In-Reply-To: Your message of Thu, 09 Apr 92 19:57:23 -0400.
Date: Thu, 09 Apr 92 21:41:34 -0400
From: John Curran <jcurran@nic.near.net>
--------
] From: tom_limoncelli@warren.mentorg.com
] Subject: Re: Settlements
] Date: Thu, 9 Apr 92 19:57:23 EDT
]
] ANS claims that they don't want to join CIX until they research this
] kind of thing. The reasoning being that if they join CIX and all the
] traffic is other CIX members floggiong ANS lines, it would be a losing
] deal. ...they still want to "research it".
]
] ...
]
] This leaves me quite confused. With a phone system it would be a lose
] because the caller pays and the callee doesn't. Too many calls in one
] direction and you're in trouble.
]
] With data networks, however, don't both sides pay? When I FTP from my
] site (connected to PSI) to another person (say, on Alternet) don't both
] "us" and "them" pay for it?
]
] Could someone explain this to me?
I will endeavor to answer your question, but it would be best to contact
the source of the claims and ask for clarification.
You ask: "With data networks, however, don't both sides pay?" Yes; you can
be fairly confident that both ends of most Internet connections pay for their
Internet access (even if it's only the communication circuit costs.)
Note that in general sites obtain their access to the Internet "in bulk",
through a single annual purchase.
When you (on PSInet) access data on Alternet, both you and the other end
have already paid for the traffic. PSI and Alternet have agreed to exchange
traffic between them without charge via the CIX interconnect. This is
regardless of the nature of your traffic.
When you exchange data with researchers at many Internet sites, a different
type of "payment" takes place. Both ends still pay to get their traffic
to an exchange point, but the exchange takes place over the NSFNET. In this
case, the NSFNET in effect "pays" for the transport of the traffic in support
of research and education.
If the traffic is not in support of research and education (i.e. commercial),
then the NSFNET will not "pay" for its transport across the NSFNET. ANS has
a subsidiary which will provide such transport, but requires the payment of
an additional annual fee. Part of this fee is set aside for improving the
network infrastructure (presumably to offset the added commercial traffic.)
Given their approach to commercial traffic, it would be only natural for
ANS to be concerned about any source of commercial traffic which did not
provide funds to insure adequate network infrastructure. A CIX connection
represents such a source of traffic. In the past, network providers have
often interconnected as "peers"; a cost recovery scheme based upon the nature
of the traffic makes this difficult unless all of the networks participate.
Note that other cost recovery models exist for commercial traffic, and each
model has benefits in certain circumstances. When providers with widely
differing policies interconnect, any differences in their recovery models
must be considered.
I've taken a number of liberties with the above explanation; apologies to
any offended party.
/John
