[2533] in Commercialization & Privatization of the Internet
Trust and the T-3 Network
daemon@ATHENA.MIT.EDU (Gordon Cook)
Sat Mar 14 02:42:13 1992
To: com-priv@psi.com
Date: 14 Mar 92 02:16:35 EST (Sat)
From: cook@tmn.com (Gordon Cook)
<<MESSAGE from>> Gordon Cook 14-MAR-92 2:16
cook@tmn
Are there really no limits on the delivery to the NSFnet community of a
functioning t-3 network by MERIT ANS? Let me make clear that by this I
mean to wonder whether there is any point out there at all beyond which,
if all 16 NSFnet nodes are not connected to the regional and t-1 backbone
ethernets and carrying meaningful data and functioning at some defined
level of up time, the NSF will get to call in its cards and say that
someone *ELSE* has a chance to use other hardware and software to
accomplish what was alledged at the hearings can be done right now?
Look at the discussion on this list. In December we heard from ANS that
the t-3 network was complete. The uninitiated might assume that meant
that all 16 nodes were passing real live data. Uh, apparently not so.
Evidently two of the 16 nodes: Nearnet, and Barrnet, and the North
Carolina state network connected as a private ANS customer were passing
live data.
After MERIT stated in a press release on March 5th that traffic from over
half of the NSF sites had been migrated to the T-3 backbone, I am told by
a reliable source that, it was asked specifically how many of the 16
backbone sites were passing live data from the regional T-1 networks to
the t-3 backbone. The reply apparently was 6 of the 16 sites. One would
think that an announcement of traffic from over half the sites
successfully migrated would mean that 9 of the 16 nodes were connected.
If you look at the message that merit sent me, of the six sites connected,
4 were connected in february. Therefore, by process of elimination in
December, the stabilization test of the "completed" t-3 backbone was done
with two of the 16 nodes - Barrnet and Nearnet. But perhaps since
Suranet, by far and away the largest regional, is now on T-3, what Merit
meant was more that half the institutions connected to the regionals were
now on the t-3 net? I have talked in the past 6 weeks to numerous
regionals. They consistently tell me that they have been pressured to
cutover to what they believe to be unreliable t-3 service but now that
parts of the T-1 are becoming unreliable they feel they have no choice.
Meanwhile at the hearings, the assertion was made that there are reliable
off the shelf t-3 routers now on the market. I have been told by a member
of this list that NetCom (do I remember the name correctly?) builds such a
router. On the other hand I thought I heard some NSF staffers sitting
next to me disagree. It would be interesting to find out why.
The committee was quite adament that the NSFnet should not be considered
as "precompetitive" or a place for technology development. This being the
case, can it be that MERIT and ANS will be allowed to continue to use the
production network as a testbed for router design for another 24 months.
Sure maybe the routers will suddenly work next week. However, haven't we
heard a very long string of "trust us"? How long under these conditions
does trust endure?
I am a newcomer to this community. People here tell me that it has grown
up successfully because of a spirit of trust and cooperation. Could it be
that that trust has become over taxed? Is this what Jim Bruce of MIT
meant when he told this list in December that he considered the foundation
on which the NSF was building the NREN to be broken?
During the past year I have heard very little about sustainable trust.
What happened? It would be ironic if pursuit of federal dollars tied up
in NREN technology were to destroy it. Did it disappear when ANS was
created to solve a commercial use problem that according to Steve's
posting today now may not be a problem any longer? The t-3 backbone was
privatized by ANS. What good has it brought to the network community? I
am hearing some very disquieting tales.
