[ih] SATNET (seismic data, Norway, UK)

Vint Cerf vint at google.com
Tue Aug 18 15:27:38 PDT 2009


The routing tables must have been artificially hacked otherwise traffic
would have gone directly to usc-isi from the prnet/arpanet gateway.

V

----- Original Message -----
From: Noel Chiappa <jnc at mercury.lcs.mit.edu>
To: internet-history at postel.org <internet-history at postel.org>;
johnnyryan1 at gmail.com <johnnyryan1 at gmail.com>
Cc: jnc at mercury.lcs.mit.edu <jnc at mercury.lcs.mit.edu>
Sent: Tue Aug 18 16:50:53 2009
Subject: Re: [ih] SATNET (seismic data, Norway, UK)

    > From: Vint Cerf <vint at google.com>

    > The configuration of the 3 net test artificially forced traffic from u
    > PRNET to ARPANET toSATNET to ARPANET again. The routing protocol in
    > ginny strazisar's gateways was distance vector and I believe would
    > handle multiple paths and backbones.

Yes, but I don't think GGP would handle that particular configuration (using
the ARPANet as both a transit network _and_ the destination). As long as the
routing thinks a given destination network is not bifurcated (the routing
world has a 'term of art' for that, but I can't recall it), it routes to the
destination network, and once there, thinks its job is done.

(I recall this problem with great clarity, because we wanted to do the same
thing with the later high-speed satellite network, when there was going to
be
a ground-station for it at BBN - send IP packets from MIT to BBN over the
ARPANet, then over the high-speed network to California, and then over the
ARPANet to the final destination. Wouldn't work. :-)

Maybe there was some hack I didn't know about (e.g tunneling ARPANet traffic
over a SATNET link), but are you sure of that configuration? Was it maybe
just PRNET to ARPANET to SATNET, or something like that?

	Noel



More information about the Internet-history mailing list