Peter probably remembers this much better than I do.
But back in the
80s, I was using ANF10 in Stockholm, and it was a network with more than
10 nodes. But I understood it as not everyone had a direct connection to
everyone. But maybe I misunderstood something? Otherwise I would try to
figure out how that was working. There were definitely KS machines on
there, along with some KLs, and I think one KI. :-)
It was not full mesh. ANF10 has routing with metrics like DECnet and RDH even
wrote a driver to talk over Ethernet between KL's.
ODEN at QZ 1099 KL10D external memory, SMP 3 DN87S hugin, munin, balder that had terminal
lines
Nadja at KTH 2020
Aurora at SU 2020
Filip at FOA3 1091 one dn87s filippa
Kicki at me 1077 3*KI10 2 dn87 on DL10 Jenjen Trasel
Katia at KTH 1050 1*KA10 1 dn87 on DL10 (dont-remember name)
Janus at QZ DN200 acting as X.25/X.29 GW (could speak ANF10 over X25 to remote sites...)
I don't remember the exact config but the front end worked if the host on DL10 or
DTE20 went down,
and somehow it was layed out so that hugin-munin-balder could talk over sync lines when
Oden was
down and they where kind of in the "center".
So if we could boot a simulted DNxxx node with simulated sync interfaces on it and the
DTE/DL10
is down, we might have a ANF10 router. DN200's booted over the sync line.
Kicki talked to Nadja and Munin on 19.2
Nadja talked to Hygin on 19.2
Minin talked to Filippa
Aurora talked to Balder
Janus talked to Balder and Minin
One of the cool things is that you can do "set host" without logging in, so
it looked like a "cloud" with terminal lines and a bunch of hosts you could
chose from.
The front end/dn87 was configured to what "host" to connect each line to when
the line became active, and if that host was down it picked one that could talk to you
and sent you there. ANF10 routing messages has a capabilities field, it tells what
it can do.
-p