[10953] in Commercialization & Privatization of the Internet
naps
daemon@ATHENA.MIT.EDU (Hans-Werner Braun)
Tue Mar 15 23:28:15 1994
From: hwb@upeksa.sdsc.edu (Hans-Werner Braun)
To: com-priv@psi.com
Date: Tue, 15 Mar 94 20:27:38 PST
> 2. On the NAP's who's idea was this ? Can't the NAP points be used to
> slow or stop the growth of small Internet providers ? I would certainly
> think so.
I think I can probably take the blame for that for the time being,
remembering starting the process when back in 1988 or so I approached
Mike StJohns (then at DCA). I am sure you must remember the ARPAnet
core model from that time and before, with all the planeteers circling
around six mailbridges. The plan of record for the NSFNET to connect
to the ARPAnet/Milnet environment was via sonnections to the ARPAnet
PSNs, and that just did not work out very well, for more than one
reason. Thinking about it I concluded that for the NSFNET the only
reasonable way to connect to the PSNs was via LAN interconnects to the
buttergates. Mike was very receptive to the idea and involved two more
people in the discussions. NASA/Ames, proud landlord for one of the
mailbridges even back then, was very interested in housing such an
NSFNET/ARPAnet/Milnet interconnect, wanted even to participate and
connect NSI, and above and beyond then even participated in the funding
for the connection to the NSS at Stanford. ARPA contributed some
equipment as well, given that there really was a perceived need for the
interconnections. So, we had suddenly four agencies together (NSF, DCA
(now DISA), DARPA (now ARPA) and NASA), soon DOE to follow, and
FIX-West (initially called FEBA-West) was born. It was really,
absolutely, positively an imperative necessity to have well architected
and well operated and managed interconnection points between the
agencies. So much that especially some mission agencies felt so
strongly about it and the need for operational stability, that it
became very hard to connect non-Fed players. Somewhere along the lines
FIX-East was born as well, by the way. Anyway, I suspect the the
somewhat perceived FIX closedness had contributed to the creation of
the CIX, a single entity with, from what I know, single point of
failure router at least then, with back hauled lines feeding into it.
Soon the CIX became as politizised as the FIX, own rules, exclusive
club, shrill in communications with people. Besides that, the single
CIX entity and the seemingly suboptimal implementation of using a level
3 router did not seem to make things well scaling. The next, in my mind
really good, thing happening was MAE-East. Alot of credit should go to
the implementers (largely Alternet and SURAnet I think (but I may be
wrong)). More distributed, and more player involvement. But that just
happened relatively recently. In any case, the need for neutral and
well architected and robust interconnections still exists. I think NSF
wanted to put enough of it in place to create a neutral area for
whoever wants to connect to it. If it works out, fine. If it doesn't,
other means would have to be found to do the same thing, or
alternatively the system will go into only semiconnected mode, at best.
Nobody is really obligated to connect to NAPs. It is just a
demilitarized zone of a neutral opportunity to interconnect. Yes, at
some cost, but nominal, and better than having to have a political
fight before being allowed to connect. In a sense, the two FIXes,
MAE-East and the CIX are NAPs, as being network interconnection points,
just like the NSF NAPs. If you don't like them, don't use them. Bob
Collett already has said that NSF is not exactly making them rich with
the NAP awards. Just let them be, and see what happens. If they are not
needed, just like any other NSF awards, funding can be reduced or
withdrawn. For the time being they are just a stability measure at a
time where the network is getting more and more commercialized.
