Path: utzoo!utgpu!jarvis.csri.toronto.edu!me!radio.astro!helios!root
From: root@helios.toronto.edu (Operator)
Newsgroups: comp.dcom.lans
Subject: Re: RG58 C/U vs ThinWire Ethernet?
Message-ID: <802@helios.toronto.edu>
Date: 15 May 89 21:57:13 GMT
References: <1026@draken.nada.kth.se>
Reply-To: sysruth@helios.physics.utoronto.ca (Ruth Milner)
Organization: University of Toronto Physics/Astronomy/CITA
Lines: 49

In article <1026@draken.nada.kth.se> ragge@nada.kth.se (Ragnar Sundblad) writes:
>
>Now people that usually know something of what they are talking about,
>although they are sellers (not of the cable), have told me that it is
>not recommendable to build long (100<l<185m) networks with RG58, I
>should use this ThinWire cable instead if I don't want to get problems
>when there are many nodes connected.
>
>(I mentioned this to a cable seller. He told me that "it had something
>to do with the timing" (RG58 = ~0.66c, ThinWire = ~0.80c).  As long
>as I don't run at the timing margins, I don't give very much for
>that (perhapse I could get 0.000001 times better performance :-).)

We use RG58U for all our thinwire Ethernet segments, and have had no
trouble with it (actually, I was under the impression that RG58 was the
cable that ThinWire, i.e. 10BASE2, was actually spec'd around). There are 
up to 20 connections on some of them, which is 2/3 of the maximum 
in the specs, and the length ranges from about 75m-185m. The error rate 
is about .01% (i.e. 1 in 10000), although it's hard to tell what proportion 
of those errors are due to various factors such as interfaces dropping 
packets, systems putting out garbage, etc. etc. Thus the real network error 
rate is probably lower than this.  I might add that the error rate on our 
coax is about an order of magnitude higher, but there are a lot more systems 
and thus there is more potential for junk.

With only 185m of cable, I can't see timing being a problem. The difference
in propagation time at .66c vs .8c over a 185m cable round trip is roughly 
400ns, which is unlikely to affect the Ethernet timing (~50us window).

Most of our thinwire segments run around the outside of a floor, i.e. along
office windowledges and through walls (mostly plaster). When the cable
is initially installed we insert a barrel connector in each place where a
system might potentially be connected (min. 1 per room even where not
planned). This usually results in about 15 connections, so there are some
extras in large offices where more than one person works. We also try to
cover the perimeter of the floor well under the length limit, so that there
is allowance for extra pieces that people may need to reach a system in 
their office when it is eventually installed. Because of the existing 
connectors, it is very quick work to connect a system later on.

Of course, we don't have to worry a lot about the aesthetics of the cabling.
Basically as long as it isn't underfoot or otherwise in the way, nobody
cares what it looks like (you should see the backs of our microVAXes!! :-) ).

-- 
 Ruth Milner          UUCP - {uunet,pyramid}!utai!helios.physics!sysruth
 Systems Manager      BITNET - sysruth@utorphys
 U. of Toronto        INTERNET - sysruth@helios.physics.utoronto.ca
  Physics/Astronomy/CITA Computing Consortium