Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!LCS.MIT.EDU!MAP
From: MAP@LCS.MIT.EDU (Michael A. Patton)
Newsgroups: comp.protocols.tcp-ip
Subject: How do you string a thinnet?
Message-ID: <8910241805.AA26176@gaak.LCS.MIT.EDU>
Date: 24 Oct 89 18:05:03 GMT
References: <1670@tfd.UUCP>
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 81


   Date: 23 Oct 89 20:08:55 GMT
   From: cucstud!tfd!kent@uunet.uu.net  (Kent Hauser)

   Beginners question:  How do you physically hook up a thinnet?
   Specifically, how do you handle multiple BNC `T' connections.

   1) Is it a bad idea to use a, say, 5-foot drop cable from the `T' to
   the host?

Extremely bad!  There should be NO drop cable, the T connector goes
right on the machine.  Ethernet is a bus network, any stubs from the
main run need to be as small as possible, much more than an inch
(including what's inside the machine, typically around 3/4 of an inch)
is probably pushing it for thin-wire Ethernet, the thick-wire spec is
a little more sensitive which is why it allows longer runs and more
hosts.

   2) Can you hang multiple hosts off in all directions from a `T'?

No, for the same reasons, there should be a T on each host and they
should be connected end-to-end, no T except at a host.

   3) Does the cabling have to be a multiple of some magic length?

No, but there is a minimum spacing which I can't recall right now (the
people who we have making cables just don't make any shorter than
this).  In practice this seldom comes into play on a thin-wire, you
probably don't want two machines that close anyway.  On thick-wire the
cable is marked with stripes at this spacing the idea being that if
you only insert transceivers at the stripes they can't be closer than
the minimum spacing.  Some people have extrapolated from this
technique to deduce that they need to be on a regular pattern, this is
not true.

   4) How can you test your network to see if you screwed something up?

Try operating it and see if it works :-).  I don't think there is any
way, after the fact, to determine that some entire installation meets
the specs.  The only way to determine this is by carefully observing
the installation and seeing that it's done right.  The technique I use
is to continually remind the people who install our cable about the
things not to do.  Whenever someone posts a message to any list I read
with rights and wrongs of Ethernet installation, I forward them a copy
(and they'll probably get this one when it comes back).  The one
technology that might be useful for after the fact testing is a TDR,
but for small installation (like you seem to be talking about below),
this might increase the cost of your network by an unacceptable amount.

   5) Or, if you've only got 6 hosts & a couple of hundred feet of RG-58,
   does it really not matter to much?

Well, when your net is that small it may not matter.  But your net
won't stay small, and if you don't obey the rules from the start you
will find someday that you add something in a legal fashion and the
net breaks.  Then your only choice is to pull it all out and start
over, doing it right.  Let me tell you, this can be quite painful.

   Now the $64k question: Is there a good reference for this kind of
   info.

Gee, you found one I don't have a good answer to.  I use lots of
references, the most important being my experience and what I've
learned talking to others (in forums like this and just standing
around in the corridors at conventions :-).  I've been hacking
Ethernet (and Ethernet-like technologies) since the late 70's and
haven't found the need recently of an overview book specifically
oriented at Ethernet and I don't seem to have such in my collection.

   Thanks much in advance.

Your welcome.  I'm sorry about missing on your last question,
hopefully someone else will chime in with some references.

	    __
  /|  /|  /|  \		Michael A. Patton, Network Manager
 / | / | /_|__/		Laboratory for Computer Science
/  |/  |/  |atton	Massachusetts Institute of Technology

Disclaimer: The opinions expressed above are a figment of the phosphor
on your screen and do not represent the views of MIT, LCS, or MAP. :-)