[87039] in North American Network Operators' Group
Re: QoS for ADSL customers
daemon@ATHENA.MIT.EDU (Frank Coluccio)
Tue Nov 29 15:41:51 2005
From: Frank Coluccio <frank@dticonsulting.com>
To: NANOG <nanog@merit.edu>
Reply-To: frank@dticonsulting.com
Date: Tue, 29 Nov 2005 14:30:03 -0600
Errors-To: owner-nanog@merit.edu
=0D
=0D
<p class=3D"MsoNormal"><span>Sean, I think your post highlights a major, ye=
t very=0D
seldom discussed, distortion of some of the more recent issues that surface=
=0D
when discussing neutrality. And these are widely shared in the industry, im=
o. IMO, you=0D
have adequately characterized the matter, as many would regard it today. Bu=
t I,=0D
along with a growing number of others no longer see neutrality in the abstr=
act.=0D
Rather, it's about choices that both end users and non-dominant competitors=
=0D
have -- or more to the point, don't have-- in invoking their own desires fo=
r=0D
isochronous-like performance, or whatever performance profile they wish, wh=
en=0D
they need it, and relegate best effort to other times OR, in fact, maintain=
=0D
only a best-effort environment, as the case may be.<br />=0D
<br />=0D
Given the unpredictable nature of end-to-end over Internet resources, focus=
=0D
generally shifts to the part of the experience that is most predictable and=
=0D
controllable. And that part is the end user's access loop section to his or=
her=0D
ISP. Beyond that point of interconnection it has always been assumed that,=
=0D
within acceptable percentile ranges, transit and core resources were more t=
han=0D
adequate to support whatever the user had to throw at it. In fact, this may=
=0D
have been true in the majority of cases in the past, but today, with speeds=
=0D
being what they are in broadband I'm sure that the tide has reversed itself=
for=0D
some domains of networking, or in some percentage of all cases. But let's=
=0D
assume that it hasn't for a moment.<br />=0D
<br />=0D
As I see the problem today, the neutrality issue has more to do with whethe=
r=0D
end users themselves are empowered to add to the print sizes in their yello=
w=0D
page ads, when they so choose. Short of this capability, at the very least=
=0D
users do not want their own applications de-prioritized by vertically=0D
integrated providers of pipe.<br />=0D
<br />=0D
Users are becoming more educated and skilled in the ways of networking all =
the=0D
time. And where they are lacking, the state of the art makes it possible=0D
through auto-config, self-healing and auto-negotiation for the end user to=
=0D
begin invoking their own preferences of infrastructure and traffic shaping=
=0D
through the use of residential gateways and home servers, for example, whic=
h=0D
capabilities would be extended even further if head end terminal gear were=
=0D
designed for such purposes, as well.<br />=0D
<br />=0D
Increasingly, I think we'll find over time that the issue over neutrality i=
s=0D
not only about the top-down perspective, but about the bottom having a say =
in=0D
the shaping of channel facilities, too.<br />=0D
<br />=0D
Frank</span></p>=0D
<p><br />=0D
<br />=0D
-------------------=0D
<br />=0D
<br />=0D
From: Sean Donelan <<a href=3D"javascript:top.opencompose('sean@donelan.=
com','','','')">sean@donelan.com</a>> Sent: Tue Nov 29 12:56: =0D
<br />=0D
<br />=0D
On Tue, 29 Nov 2005, Kim Onnel wrote:=0D
<br />=0D
> The links are now almost always fully utilized, we want to do some QoS=
to=0D
<br />=0D
> cap our ADSL downstream, to give room for the Corp. customers traffic =
to=0D
<br />=0D
> flow without pain.=0D
<br />=0D
<br />=0D
While some people will cry network neutrality and think the Yellow Pages=0D
<br />=0D
must sell only one size listing, some people are willing to pay for=0D
<br />=0D
differentiated service. Trying to classify "bad" traffic can be=
=0D
<br />=0D
done using products like Sandvine. But it may be easier to classify "p=
remium"=0D
<br />=0D
traffic and mark it for special handling, and then treating everything=0D
<br />=0D
that isn't marked as premium traffic as best effort traffic.=0D
<br />=0D
<br />=0D
But expect great wailing and gnashing of teeth over setting or changing=0D
<br />=0D
DSCP/TOS bits or creating different queues for different traffic. Should=0D
<br />=0D
DSCP bits in IP headers be treated like TTL bits which are modified by=0D
<br />=0D
the network. Should ISPs use anti-spoofing techniques similar to prevent=0D
<br />=0D
the use of arbitrary IP addresses to control DSCP/TOS values in packet=0D
<br />=0D
headers?=0D
<br />=0D
<br />=0D
Most routers already give priority to some types of traffic, such as=0D
<br />=0D
routing update packets.=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
Frank A. Coluccio=0D
<br />=0D
DTI Consulting Inc.=0D
<br />=0D
212-587-8150 Office=0D
<br />=0D
347-526-6788 Mobile=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
<br />=0D
</p><BR>=
