NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/dcom/modems-part1/847

Text File | 1996-08-05 | 3.1 KB | 64 lines

Newsgroups: comp.dcom.modems
Path: istar.net!infoshare!whome!gts!bokonon!stephen
From: [e-mail] (Stephen M. Dunn)
Subject: Re: Is UUCP is critical feature for Unix machine?
Organization: United System Solutions Inc.
Date: Mon, 8 Jan 1996 03:59:47 GMT
Message-ID: <[e-mail]>
References: <4cce5p$[e-mail]>

In article <4cce5p$[e-mail]> [e-mail] (Salem Reyen) writes:
$Currently, there are only two brands of modems (MultiTech and Microcom)
$provides the UUCP feature for UNIX.  Is this a "must-have" feature for a modem
$using under Unix environment?  How useful is it in real life?

   Telebit provides this, too.

   The feature to which you refer is called UUCP spoofing and generally
works with the UUCP-g protocol.  UUCP-g is commonly implemented
using 64-byte packets and a window size of 3, though it's certainly
possible to implement it with other settings.  As the headers are
6 bytes, that means there can be no more than 210 bytes outstanding
before the sending side stops sending and waits for an acknowledgement.

   For various reasons, this may cause throughput problems.  On
half-duplex modems, for example, the time to turn the link around
may cause serious throughput degradation.  Even for modulations
such as HST, in some situations the link may reverse to handle
the UUCP acks (I've seen this happen, giving ~300 cps over an
HST link with a forward data rate of 14.4 kbps).  If you have
data compression and/or error correction enabled, the delays
involved may interfere with decent UUCP throughput as well.

   Therefore, some modems perform spoofing.  The modem on the
sending side will send an acknowledgement to the sending machine,
even though the data has not yet reached the far side, and it
will take care of ensuring that the data is transmitted
to the other modem.  The acknowledgements from the remote
machine are silently swallowed.  Basically, this transforms
one end-to-end link into three separate links, so that the
UUCP protocol no longer has to worry about round-trip delays
over the entire link.

   UUCP spoofing on Telebits, using variations on the PEP
protocol, is a very good thing, as PEP (at least the original
one) is half-duplex.  I'm not sure about TurboPEP.

   For full-duplex modulations with reasonable symmetry,
such as V.34, V.32bis, V.32, etc., you will probably get
some benefit from spoofing if you are also using error
correction.  I know that even at g(64,7), I don't get the
full theoretical bandwidth out of a non-spoofed V.32bis
connection - though I suspect there may be delays on one
or both computers in many cases which also decrease throughput.

   If you're not passing much UUCP traffic, though, then
UUCP spoofing won't really give you much of a benefit.  And
if you're not using UUCP-g, chances are your modem won't
spoof whatever UUCP protocol you _are_ using.
--
[e-mail]               ...!{xrtll,gts.org}!bokonon!stephen
----------------------------------------------------------------------------
Stephen M. Dunn, CNE, ACE, Sr. Systems Analyst, United System Solutions Inc.
104 Carnforth Road, Toronto, ON, Canada M4A 2K7          [phone] x251