Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!rutgers!brl-adm!seismo!sundc!nears!occrsh!occrsh.UUCP!tiger.UUCP!authorplaceholder
From: authorplaceholder@tiger.UUCP.UUCP
Newsgroups: comp.dcom.modems
Subject: Re: Documentation for MNP
Message-ID: <144100001@tiger.UUCP>
Date: Sun, 8-Feb-87 10:33:00 EST
Article-I.D.: tiger.144100001
Posted: Sun Feb  8 10:33:00 1987
Date-Received: Tue, 10-Feb-87 03:03:47 EST
References: <272@gaia.UUCP>
Lines: 55
Nf-ID: #R:gaia.UUCP:-27200:tiger.UUCP:144100001:000:3045
Nf-From: tiger.UUCP!rvk    Feb  8 09:33:00 1987


/* Written  8:36 pm  Feb  1, 1987 by zhahai@gaia.UUCP in tiger.UUCP:comp.dcom.mode */
/* ---------- "Documentation for MNP" ---------- */
I gather that MNP is Microcom Networking Protocol, and is a error
correcting protocol or set of protocols, and is generally available
in modems if anywhere (ie: in a few modems, but not usually in
terminal or host software).  Could someone tell us a little more
about it?  Is it proprietary, and how much so ($) if yes?  Is it
implemented at the async character level or at the internal synchronous
level in 212/v.22/bis type modems, etc?  Is it forward error correcting
or automatic retry request?  How much overhead is there?  What is the
packet size, and how does it decide to send a short packet (if
it is packet oriented... I am guessing obviously).  Where can
documentation be found and at what cost?  Also, how well does it work
(anybody using it?)?   

I suspect there would be sufficient interest to justify posting good
info to the net; if I am wrong, such as if this has already been covered
in detail before I started listening to this group, please let me know. ~z~

-- 
Zhahai Stewart
{hao | nbires}!gaia!zhahai
/* End of text from tiger.UUCP:comp.dcom.mode */
Zhahai,

I own some Microcom modems but Microcom doesn't tell its users much.  The
protocol is definetly modem resident, hence, running another block mode
error correcting protocol (e.g. UUCP) over it will further increase over
head.  With the serial port set to 9600 bps ( it can be set to 19.2 Kbps
although the modem reportedly actually transmits about 11Kbps) the modem
transmits 256 byte packets of user data in half duplex.  UUCP throughput
is about 500 characters per second (5000 bps) since there is double 
overhead here.  UUCP actually gets about 7000 bps then throws on its own
overhead.  Although the modems have dozens of software options and switches
I have enjoyed using them.  I tested Fastlink and hated them.  There down
fall is a too large packet size and timeout.  When using them for terminal
traffic they wait about 2 seconds for the buffer to fill before sending.
Then there is another 2 second delay at the other end for the echo.  This
made terminal work very frustrating.  The same thing killed UUCP.  Since
a UUCP block is apparently too small to fill the buffer, a 2 second delay
is incurred before each block is sent, then 2 seconds pass waiting for the
"ACK".  UUCP throughput was a sickening 633 bits per second.

Back to the subject.  I think PC magazine recently had an article about
high speed dial up modems.  Fastlink came out better on clean phone lines.
Microcom came out better on noisy ones.  I can vouch for the latter.
My modem simply error corrects over Call Waiting Beeps on my line.

If you want to run an error correcting protocol from host to host, better
go with a none error correcting modem and run KERMIT or X.PC.  MNP is 
proprietary and resides from modem to modem, which helps terminal traffic
but is bad for running already blocked file transfers.

Ray Kellogg ihnp4!ocsmb!rvk