Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!uw-beaver!blake!oregon!jqj
From: jqj@oregon.uoregon.edu (JQ Johnson)
Newsgroups: comp.dcom.modems
Subject: Re: V.32 @ 38,400 bps MNP Class 9 Microcom Modem? Is this for real???
Message-ID: <1318@oregon.uoregon.edu>
Date: 28 Apr 89 17:58:31 GMT
References: <28111@cci632.UUCP> <118@dsinc.DSI.COM>
Organization: University of Oregon
Lines: 13

In article <118@dsinc.DSI.COM>, syd@dsinc.DSI.COM (Syd Weinstein) writes:
> Thus for uucp, and compressed news transmission, the tb+ is still
> a higher throughput modem.  About 1.3 to 1.4 times faster.
> 
> For restricted character set ascii, the mnp9 v.32 will be faster,
> but how much data do you send that way and how much is compressed?
Has anyone done any evaluation of v.32 + MNP-9 for TCP/IP SLIP traffic?
I am interested in thruput for large packets, but also in behavior for
small packets (e.g. 60 to 150 bytes typical of telnet traffic).  In that
realm, most of the packet is binary header and is mostly the same from
packet to packet.  We care less about throughput than about latency:
a modem that takes 50ms to do the compress/mod/demod/decompress adds
100ms to RTT on a link.