Path: utzoo!attcan!uunet!lll-winken!lll-ncis!helios.ee.lbl.gov!pasteur!ucbvax!ulysses!smb
From: smb@ulysses.homer.nj.att.com (Steven M. Bellovin)
Newsgroups: comp.dcom.lans
Subject: Re: Ethernet line extenders experiences?
Keywords: ELE Ethernet transceivercable
Message-ID: <11103@ulysses.homer.nj.att.com>
Date: 15 Jan 89 14:51:08 GMT
References: <582@philtis.UUCP>
Organization: AT&T Bell Laboratories, Murray Hill
Lines: 20

The problem with Ethernet line extenders is that you have to be careful
of your delay budget.  That is, in an Ethernet configured according
to the letter of the spec, each component -- coax, repeaters, drop cables,
etc. -- has been assigned a delay time.  The sum of the maximum possible
end-to-end delays must be less than a certain amount, to wit the round trip
time for a minimum length packet.  If you increase the length of a drop
cable, you must decrease the maximum coax length that you use.  You must
also allow for propagation delay through the extenders.

Let me give some numbers.  The signal speed in thick (or for that matter, thin)
coax is given in the spec as .6667 c, whereas the signal speed in the
transceiver cable is .5 c.  Thus, an increase in drop cable length from
50 to 250 meters must be compensated for by a decrease in the maximum
diameter of the net by 267 meters, thus *decreasing* the maximum extent.
If you're careful, and if you know what you're doing -- for example,
only putting long drop cables in the middle of the net, or restricting
other components -- you can get away with it.  The vendor may be able to
help you as well.

Bottom line -- you can sometimes use such boxes.