Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!sharkey!atanasoff!shaver
From: shaver@atanasoff.cs.iastate.edu (Dave Shaver)
Newsgroups: comp.dcom.lans
Subject: Re: Cabletron Twisted Pair Ethernet
Message-ID: <927@atanasoff.cs.iastate.edu>
Date: 30 Mar 89 00:25:21 GMT
References: <1297@cunixc.cc.columbia.edu> <28703@bu-cs.BU.EDU>
Reply-To: shaver@atanasoff.cs.iastate.edu (Dave Shaver)
Distribution: usa
Organization: Iowa State U. Computer Science Department, Ames, IA
Lines: 69

Kent England writes:
>As many of you know, I like Cabletron very much, but I think
>that all the TP Ethernet products are working well and will even
>interoperate when the 10BaseT is finished [...]

We're (finally) moving to TP for our LANs on campus.  Much has been
said on the TP Ethernet issue here, but I've never really see a summary
of the "whole" issue in any of the postings.  Thus, I'm interested in
people responding via mail and I will post a summary of the following
issues and any other comments or experiences you would like to add:

First, with most of the TP solutions (Cabletron, Synoptics, etc) there
is one or more boxes that tie all the TP segments together (for
Cabletron it's the MMAC.)  A basic question we have is: What about
backplane contention?  Will this be a problem?  e.g.: Let's say we have
several diskless workstations on their own TP segment and the server on
another.  Assume that the server and diskless stations can be on
different cards (in the Cabletron example TPT-MIMs.) Is this "slower"
in any important way than if the two were connected via coax?
Cabletron reps and techies claim no.  We're looking for someone with
experience.

Next, we currently have a campus-wide "network" (read terminal server)
which requires 2 pairs for a connection.  However, each "network" jack
on Campus has 3 pairs.  Thus, a user can use the "network" and still
have a pair left over.  At least one department on campus has gone with
Pairtamer (I think the same as a Baluns) for TP Ethernet.  The
Pairtamer takes RG58 and transmits it over a single pair.  3Com's sales
blurbs say the segement of TP works exactly like coax, so you can do
something like this without any problems:

X = server
x = client
z = random non-client
O = Pairtamer
= = Coax
- = Single TP

=z=X===x==O------O===x==z==O-----O====x===z===O----O=z
			--or--
MM = Multiconnect

	z=O--------O=||
		     ||
       z=x=O-------O=MM=O--------O=X
		     ||
		     ||=O--------O=x==x==x

I'm wondering if someone could relate their experiences with
Baluns/Pairtamer vs.  a "real" TP solution (Cabletron, Synoptics,
etc.)  All of the "real" solutions I've seen require 2 pair.  In
comparing the use of a "real" 2 pair solution vs. using a single pair
solution (Baluns), is there enough functionality, reliability, etc,
gained to warrant the major expense (on our campus) of pulling more
wire to each endpoint?  Cabletron reps say 2 pair solutions are the
only way to go.  Their major selling point seems to be in the realm
of network management.  Comments PLEASE.

Finally, more experiences and comments (like Kent's) with either
Cabletron, Synoptics, 3-Com Multiconnect, etc, are highly sought
after.

All of us neyphites in the TP Ethernet game thank you in advance.

/\  Dave Shaver  -=*=-  CS Systems Support Group, Iowa State University
\\  UUCP:  {hplabs!hp-lsd, uunet!umix!sharkey}!atanasoff!shaver
\/  Internet: shaver@atanasoff.cs.iastate.edu

		...In stereo where available...