[rescue] AUI duplex [was Re: New SPARCStation 10 owner, needs help with some issues though.]

[Mouse]

> Once some of the snow melts, I'll see if I can scavenge up enough
> bits to commission a test length of 10-Base-5 (good old-fashioned
> "Etherhose") and see what various machines can do with it.

I'll be interested.  I no longer have any 10base5 stuff, though I could
probably set up a small 10base2 network.  The only remnants of 10base5
I still have are things like AUI "multiport transceivers" that really
aren't 10base5 per se, but instead AUI stuff dating from 10base5 days.

> Using a switch in the middle stretches the validity of the test a
> little bit.

Well, sort of.  It just makes it look, from an original-Ethernet
perspective, like two segments with a higher-layer (ie, packet-layer
instead of carrier-and-media-layer) device in between.  (Everything
except the SS20<->Catalyst link is also at least an order of magnitude
faster than that link, too.)

Would it make you feel better if I used two machines connected directly
with a crossover 10baseT cable? :-)  (I have the pieces to put together
a 10base2 network, but much of them are I-don't-know-where in storage.)

> Recall that before transmitting, a machine had to first listen to see
> if it was safe to "talk".

...and then, if it sounded quiet, transmit channel seizure and watch
for collisions.  (I always wondered what happened if two stations an
integral number of bits apart started channel seizure at the same time.
Late collision, perhaps?)

> That takes time; and if one adds collisions into the mix things can
> stretch out quite quickly with the result being less bandwidth that
> actually gets used.

Certainly.  But on an otherwise quiet network, two machines speaking
TCP to one another should approximately never collide, reaching close
to medium bandwidth limits.  Using gear that has collision lights I've
seen that happen: transmit lights for two stations on solid, with the
collision light only barely flickering every second or two.

> NFS-over-UDP can get so unreliable [...]

Oh, it can, it can indeed.  For much of my career I was co-admin of a
network at a university where we used NFS-over-UDP heavily.  I've seen
plenty of rather interesting failure modes.  (I think the most
interesting was one where NFS reads of files would hang when the file
size fell into one of certain small ranges.  Turned out to be a minimum
packet size issue: one host was generating packets small enough that
another host would drop them as runts.  UDP overhead meant this
happened only with last frags, hence the odd file size dependency.
Remounting with rsize=1024 made the symptom go away, as it meant
IP-layer fragmentation never happened.  I never found a reference
authoritative enough to tell me which host was de-jure at fault.)

> Don't let a vague similarity at the user layer fool you.  It's not
> about coax any longer, and hasn't been for decades.  It's all
> crossbar-switches and connection-oriented, [...]

Connection-oriented?  Are you talking about flow labeling?  That's
relatively recent, isn't it?

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse at rodents-montreal.org
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B