[ previous ] [ next ] [ threads ]
 
 From:  David W. Hess <dwhess at banishedsouls dot org>
 To:  m0n0wall at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall] Cable modem incoming pipe bandwidth
 Date:  Wed, 31 Jan 2007 15:47:38 -0600
On Tue, 30 Jan 2007 23:25:34 -0500, "Chris Buechler" <cbuechler at gmail dot com>
wrote:

>Cable modems have a fixed speed cap. Whether or not you can actually
>utilize that to its full capacity depends mainly upon congestion in
>any portion of your ISP's network and the speed of the remote server
>(or the connection between the remote server and your ISP). It's
>unlikely, unless your ISP's network is severely mismanaged, that your
>actual connection is varying from 4-11 Mb. The reality is likely that
>whatever you're connecting to can only reach that speed at that given
>time. Another possibility is traffic shaping of some sort on your
>ISP's network.

I would not put it past a broadband ISP to have these types of problems but they
do seem rare.  Back when I had cable, I had similar issues where we verified
limited throughput below the link speed which varied over a 24 hour period
caused by congestion close on the ISP side but at the time I was not in a
position to do traffic shaping anyway outside of what the applications I was
using supported.

>You should be able to set that to the actual cap and be fine. If your
>ISP's network routinely gets bogged down and you can't actually reach
>your cap, there isn't anything you can do about it. It would be
>theoretically possible, though difficult, to write something to detect
>changes in your actual maximum achievable throughput in near real time
>and change pipes accordingly, but I don't know of anything that
>permits something of that nature.

I have not had an occasion to try it yet but several people have setup the BSD
or Linux traffic shaping facilities with a monitoring script or program that
measures latency to a close router on the ISP side of the DSL or Cable link and
adjusts the maximum traffic shaper throughput accordingly.  I suspect this would
work rather poorly in some circumstances without sophisticated tuning unless the
ISP enforced some type of equal sharing at the point of congestion.