🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: crl.general
Article #52

Re: Interconnect speed

#52
From: Matt Dillon
Date: Thu, 08 Dec 1994 08:03
56 lines
3771 bytes
:In article <3c6bkg$9d3@gap.cco.caltech.edu>,
:Tatsuya Murase <tots@wrath.ugcs.caltech.edu> wrote:
:>stupid@crl.com (Eric Penn) writes:
:>
:>>I've seen some people posting output from a program that seems to be called
:>>"traceroute" that shows the routing and speed of a connaction from thier
:netrek.  ping times of less than 100ms are very nice, but if you are
:using a modem, people say that the minimum is around 300ms.  Ping also
:lists the percentage of packets that don't make the round trip ("dropped
:packets", "packet loss").  It also gives the average round trip time, the
:standard deviation, and the minimum/maximum round trip times.
:
:--tots

    A high speed V.34 asynchronous modem will yield on the order of 150ms ping
    time (standard ping with 56 data bytes) if you have the local
    modem<->host links running as fast as possible.

    Theoretically, you would think that a 64 byte packet on a 28.8Kbps
    link would take only 44mS round trip.  There are several factors that
    increase round trip delays.

    A reliable connection 'packetizes' data in the modem before
    modulating it, and the data is sent out over the wire in packets.
    This creates additional delay and causes burstiness.  In order to
    minimize round trip time, you want to set the local host<->modem
    link to run as fast as your computer can reliably handle.  After all,
    a local host<->modem link at 28.8Kbps would add another 44mS.  If
    the modem on the other end has a local link speed of 28.8Kbps, that's
    yet ANOTHER 44mS, yielding 44x3 = 132mS minimum theoretical round
    trip time **for a small packet**.  (note that modems not in reliable
    mode do not packetize data and may result in a quicker round-trip
    because the characters are pipelined, so the local link speed does
    not add significant delay.  However, you then do not get end-to-end
    flow control or an error-free connection).

    On the otherhand, if you run the local host<->modem link at, say, 
    115Kbps on both ends, you wind up with 44 + 2 or 46mS minimum
    round trip time.

    Portmaster's can generally go that fast, but PCs are usually
    limited in how fast you can run their serial ports reliably,
    especially in the face of Windows, OS/2, or some other 
    operating system.

    There are also other delays... compression adds delay (but reduces
    over-the-wire packet size and so may reduce the delay overall), modem
    buffering adds delay, the underlying modulation adds delay, and the
    reliable protocol packetization adds delay.

					-Matt

-- 
    Matthew Dillon	director, BEST Internet Communications, Inc.
		    <dillon@best.com>, <dillon@apollo.west.oic.com>
    [always include a portion of the original email in any response!]
Technical Details
Message-ID: <3c6eku$pb9@shell1.best.com>
Path: arc1-ca.pugleaf.net!bga.com!news.sprintlink.net!hookup!newshost.marcam.com!news.kei.com!ddsw1!news1.best.com!shell1.best.com!not-for-mail
References: <3c5m64$584@crl10.crl.com> <3c6bkg$9d3@gap.cco.caltech.edu>

Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 1120a5,a9223f802e9476af
X-Google-Attributes: gid1120a5,public
X-Google-Thread: 115608,a9223f802e9476af
X-Google-Attributes: gid115608,public
X-Google-Thread: f4e7e,a9223f802e9476af
X-Google-Attributes: gidf4e7e,public
X-Google-Thread: fa874,a9223f802e9476af
X-Google-Attributes: gidfa874,public
X-Google-ArrivalTime: 1994-12-08 04:07:02 PST
Path: arc1-ca.pugleaf.net!bga.com!news.sprintlink.net!hookup!newshost.marcam.com!news.kei.com!ddsw1!news1.best.com!shell1.best.com!not-for-mail
From: dillon@best.com (Matt Dillon)
Newsgroups: crl.general,ba.internet,comp.sys.amiga.datacomm,rec.games.netrek
Subject: Re: Interconnect speed
Date: 8 Dec 1994 00:03:25 -0800
Organization: Best Internet Communications, Inc. (info@best.com)
Lines: 56
Message-ID: <3c6eku$pb9@shell1.best.com>
References: <3c5m64$584@crl10.crl.com> <3c6bkg$9d3@gap.cco.caltech.edu>
NNTP-Posting-Host: shell1.best.com