Prickle-Prickle, the 26 day of Discord in the YOLD 3186
How Internet came to Sweden
Originally published 2020-04-08 20:45 +0200. Republished with links to videos instead of just embedding them so the people in the feed can follow them. Republished again 2020-09-28 when Invidious had closed down.
Looking through some old RFCs I came across this in RFC 900 from June 1984:
R 192.005.050.rrr CTH-CS-NET Chalmers CSN Local Net [93,UXB]
This is the first mention I have found of IP addresses allocated for something in Sweden.
If we consider "having an IP address" as "being on the Internet" this is probably how Internet came to Sweden.
"Chalmers" is, of course, Chalmers Institute of Technology in Gothenburg, Sweden. The IP range still seems to belong to Chalmers. WHOIS says:
inetnum: 192.5.50.0 - 192.5.50.255 netname: CTH-CS-NET descr: Chalmers University of Technology descr: Gothenburg, Sweden
and at least the reverse DNS says so too:
tp2:~ % host 192.5.50.1 1.50.5.192.in-addr.arpa domain name pointer cth-192-5-50-gw.chalmers.se. tp2:~ %
"UXB" mentioned above is:
[UXB] Ulf Bilting CHALMERS bilting@purdue.ARPA
Some kind of gateway is also mentioned:
This section lists the mapping between the Internet Addresses and the
Public Data Network Addresses (X.121).
...
Internet Public Data Net Description References
--------------- ----------------- ----------- ----------
...
014.000.000.011 2405-015-50300 00 CHALMERS [UXB]
A reasonable guess is that this was an X.25 gateway for the Chalmers network.
I reached out to Ulf Bilting to find out more about this. He graciously answered a number of questions from me. There's also this interview the Swedish Internet Foundation did with Ulf for their series about the history of the Internet but it's rather shallow on the technical details:
https://youtube.com/watch?v=Fkyvap_RXUc
The actual IP gateway at the Chalmers end was a VAX 11/780 running 4.2BSD. The X.25 connection was made through a small PDP-11 running SNIP (Small Network Interface Processor) developed by a department (Teletrafik & Datorsystem) at KTH, the Royal Institute of Technology, in Stockholm. The SNIP was connected to the VAX through a parallel interface card.
A user process Ulf wrote himself in C was the endpoint on the VAX. This process used a raw socket directly towards the BSD kernel. It used encapsulation according to RFC 877 to send IP packets over X.25 and connected to the SNIP, which dealt with the actual X.25 traffic. The other end of the IP-over-X.25 tunnel was Purdue University, Indiana, USA on the CSNET.
The X.25 network the SNIP was connected to was an early attempt by the Swedish University Network (SUNET) to connect universities in Sweden. As far as I understand this was the Swedish state authority Televerket's X.25 network. At the start of the SUNET project this was mainly used to connect terminal servers, so you could dial in to your local terminal server and access a host elsewhere without having to dial expensive long distance calls.
It's a little unclear which one of Televerket's networks were used. Early on Televerket had a network called Telepak (sometimes Databas 300), which I believe was Tymnet/McDonnel Douglas hardware and had a connection to the US part of Tymnet. The connections from SUNET was probably through this network. Later generations were called Datapak I and II and used different network hardware.
The throughput from Chalmers to the US was rather low, about ~2 kilobits/second, likely because of both tunnelling overhead, the poor fit of TCP over X.25, and the fact that the Tymnet connection to the US was shared with all the other Televerket users.
Jacob Palme keeps some texts available from the early KOM conference system, basically an early multi-user BBS, at:
https://people.dsv.su.se/~jpalme/qzkom/
There's a conference called "Sunet erfarenhetsutbyte" which gives some insights on the early Sunet network. Here's a post by Ulf:
(Text 110099) 83-07-12 08.46 Ulf Bilting CTH info.beh. Mottagare: SUNET 'Swedish University Network' Sändare: Ulf Bilting CTH info.beh. -- Sänt: 83-07-12 08.46 Ärende: CSNET, goda nyheter. Vid ett möte igår i Oslo beslutades att upprätta en testförbindelse mellan Csnet och Sunet. Den kommer att installeras i åiGöteborg och kommunicera via x25 till en gateway i Csnet. Tjänsten som erbjuds är post (mail som det heter) till samliga Csnetmedlemmar och samliga noder på Arpanätet. Vi kommer att tillåtas att upprätta en egen gateway för vidare distribution inom Sunet.
which talks about an early test connection, initially for just mail, between SUNET and CSNET, probably what turned into the Chalmers site which had real IP connectivity.
It's interesting that the tunnel was between Chalmers in Gothenburg on the Swedish west coast and Indiana, USA, instead of something closer… For instance, NORSAR, the Norwegian Seismic Array, in Kjeller outside of Oslo is much closer to Gothenburg and had ARPANET connectivity since 1973! In fact, the NORSAR satellite uplink was Tanum in Sweden, just to the north of Gothenburg!
The real reason to connect to CSNET instead of using the uplink in Tanum was probably political rather than technical.
Note that the rest of SUNET didn't get access to the Internet just because parts of Chalmers did. As far as I understand an official connection to Internet for the entire SUNET didn't happen until 1988.
Fun fact: Ulf is an early computer musician. He has a number of noise compositions on Youtube.
Ulf Bilting: "Two phrases":
https://youtube.com/watch?v=QWWo_4qwAHY
With Zbigniew Karkowski: