From owner-freebsd-hackers  Mon Jun 26 16:49:00 1995
Return-Path: hackers-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id QAA21958
          for hackers-outgoing; Mon, 26 Jun 1995 16:49:00 -0700
Received: from mail.htp.com (mail.htp.com [199.171.4.2])
          by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id QAA21948
          for <freebsd-hackers@freebsd.org>; Mon, 26 Jun 1995 16:48:55 -0700
Received: from et.htp.com (et.htp.com [199.171.4.228]) by mail.htp.com (8.6.5/8.6.5) with SMTP id TAA06713 for <freebsd-hackers@freebsd.org>; Mon, 26 Jun 1995 19:47:41 -0400
Date: Mon, 26 Jun 1995 19:47:41 -0400
Message-Id: <199506262347.TAA06713@mail.htp.com>
X-Sender: dennis@mail.htp.com
X-Mailer: Windows Eudora Version 2.0.3
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
To: freebsd-hackers@freebsd.org
From: dennis@et.htp.com (dennis)
Subject: Re: FreeBSD as a router 
Sender: hackers-owner@freebsd.org
Precedence: bulk

>: latency is a factor of throughput. Throughput is a timed measurement...more
>: latency, less throughput.
>
>No.  That is not the case.  If I keep the pipe full, then I can have a
>relatively large latency, but still get good throughput.
>
>For example, I routinely get 2.5K/s from a machine that is two 28.8 kbps
>modems from the internet.  The interactive latency is high, but I'm
>still able to keep the pipe full of data.  I also get 2.5K/s one hop
>away as well.  If what you are saying is true, I should get no better
>than 14.4kbps, but I get closer to 28.8kbps.
>
>If you double latency, you do not decrease the throughput by 1/2, but
>you need to double the window size because the "energy" of the pipe is
>twice as great as the lower latency case.
>
You (and many others) are confusing throughput with utilization. Throughput
is a performance measurement. Keeping a pipe full because of backlog does
not equate to 100% throughput....

db