[AusNOG] [LINK] [ISOC-AU-mems] Happy Birthday ... AARNet

[Geoff Huston]

>>
>
> I specifically remember a slip connection to Hawaii growing from 1200
> bps to 2400 bps preceeding the 56Kb frame relay connection.

Nothing to do with me, and I have no knowledge of it. The first  
connection in this setup with the collaboration with the folk at the  
University of Hawaii was a 56K digital circuit using a hemispherical  
transponder on an Intelsat spacecraft with earth stations in Hawaii  
and Australia (Oxford Falls as I recall) and circuit terminations at  
the University of Hawaii (Dr Torben Nelisen) and the University of  
Melbourne (Robert Elz), all fully funded by NASA's Paccom program.


> And I first remember the 2400 Baud connection circa 1985 and I think  
> it
> was driven by Phil Mcrea from Ansto and paid for by CSIRO.


None of this ties in with my recollection at the time.

> I believe that Melb, Syd universities connected to Ansto to obtain  
> their
> feeds.

Again none of this ties in with my recollection at the time.

The earlier work used message relay systems. There were a few of them  
but the one that I was familiar with was managed by Robert Elz of the  
Computer Science Department of the University of Melbourne. At some  
time he had funding support from Digital Equipment, but this was for a  
limited period and then the Australian community served by this mail  
relay went into a user pays regime where Robert sent out the bills  
periodically (monthly?). It was a uucp connection across conventional  
modems, and the speed was what ever the modems could converge onto.  
(diversion of the issue with mu-law / A-law conversion of audio  
circuits at the US IXCs and their impact on modem speeds abbreviated).  
He worked with Rick Adams, who operated a relay system at seismo that  
eventually morphed into the uunet juggernaut andf Piers Bertma at CWI  
in the Netherlands and I suspect that he also used one or two others.  
Within Australia there was pretty widespread use of ACSnet protocols.  
The advantage of this system over uucp was the use of 3 virtual  
circuits in each direction, allowing small messages to 'overtake'  
larger messages across the network.ACSnet was maintained by Piers  
Lauder at the University of Sydney, supported by Bob Kummerfeld, and  
most of us ran local relays and gateways into various other systems  
and networks. Australian reticulation was done by whatever was at hand  
- for a while the community piggybacked on the csironet X.25 network  
because of some anomalies in their charging system allowed this form  
of edge-to-edge traffic to be supported at very low cost, but in the  
late 80's a number of places started doing inter-capital leased lines  
(Sydney to Melbourne first, then in 1989 Adelaide to Melbourne and  
Canberra to Melbourne). The first of these leased lines was 9.6k (syd  
- melb) and the later ones ran at 48K (which were nailed up 64K  
digital voice carriers from the telco fabric with 2 bits per frame  
ripped out by encoding, clocking and CPE O&M for the DDS service).

Yawn. That was years ago. On to today's problems. What are we going to  
do given that noone is doing anything remotely serious in IPv6 and the  
crunch time of IPv4 address exhaustion is getting ever closer? If we  
can't manage to preserve some level of protocol coherence across the  
network in the coming few years then we may end up not much better off  
than the situation on 20 years ago. Or do we say goodbye to all this  
end-to-end IP stuff and just run client sever over http and forget  
than anything else was ever possible?

g