16 Oct
2025
16 Oct
'25
12:44 p.m.
Paul, can you put that document up somewhere? It would be an interesting read.
Sent from my iPhone
On Oct 16, 2025, at 07:12, Paul Koning via cctalk
<cctalk(a)classiccmp.org> wrote:
On Oct 15, 2025, at 7:14 PM, Doug Jackson via
cctalk <cctalk(a)classiccmp.org> wrote:
Ethernet also got *way* more market traction, because it was
infinitely more survivable.
One of my early jobs was managing a Token Ring network, and we spent our
days running around a 13 floor building, chasing machines where people had
kicked connectors out of walls, just enough to stop data, but not enough to
make the MAU do the self isolation. We had a piece of software called the
Cabletron TR Manager - that monitored the ring for beaconing, and let us
know the upstream node that detected the break. Then we would consult our
*detailed* notes on what cards were installed where, so we could find the
culprit that was broken. Without the notes, we would have had zero chance.
FDDI was somewhat better. It's not all that well known, but IBM (802.5) token ring
and FDDI have essentially nothing in common. At DEC when we were working on its
development we liked to say that the only things in common are "token" and
"ring". Actually, FDDI is in a sense an evolution of 802.4, of all things.
I also remember while there kicking around the notion that we could take the FDDI
signaling scheme (4b/5b code) and use it to send Ethernet packets. That worked quite well
and the rest is history...
Heady days.
I suggested to the network manager at the time that we could transition
from TR to Ethernet (everything was wired with Cat3 Shielded cable - but he
didn't want to, because "Ethernet had collisions" - that was when I
discovered that everybody has limitations that something breaks their
thinking. After a while I convinced him to transition one of the Cabletron
cards to Ethernet, and do a test on a 32 workstation card - Suffice to say
that those 32 machines never had an issue, and eventually, all 800 machines
across two rings were transitioned to 100Mb Ethernet.
Around that time, IBM put out a marketing document that pretended to show why token ring
was better than Ethernet. The DECnet architecture group (where I worked at the time)
created a paragraph by paragraph rebuttal to that and published it as a joint DEC/3Com
document. I still have it: "The Digital/3Com Guide to IBM document #
GG22-9422-0" (DEC document EE-EA345-42-001). It doesn't seem to be online,
Google has never heard of it, nor the IBM document it rebuts. As I recall, Bill Hawe was
the lead author of that work; I wrote some bits and pieces for it but I don't remember
the details.
paul