[171901] in North American Network Operators' Group
Re: Observations of an Internet Middleman (Level3)
daemon@ATHENA.MIT.EDU (Vinny Abello)
Fri May 16 09:10:31 2014
X-Original-To: nanog@nanog.org
Date: Fri, 16 May 2014 07:56:25 -0400
From: Vinny Abello <vinny@abellohome.net>
In-reply-to: <CF9B77D1.D1980%jason_livingood@cable.comcast.com>
To: "Livingood, Jason" <Jason_Livingood@cable.comcast.com>
Cc: nanog@nanog.org
Errors-To: nanog-bounces@nanog.org
On , Livingood, Jason wrote:
> On 5/15/14, 12:49 PM, "arvindersingh@mail2tor.com"
> <arvindersingh@mail2tor.com> wrote:
>
>
>> I have two issues with the comments:
>>
>> 2. You mention that all packets treated equally - no games. Why does
>> AS7922 assign the speed test different DSCP from regular internet
>> connection?
>
> I have no idea what you are talking about. Our Internet traffic,
> including
> to speedtest web sites, is all best effort class data. Do you have more
> specific information?
>
> Jason
I think he's questioning why packets from speedtest.comcast.net have CS1
if everything is supposedly equal, and what that is used for. A quick
Wireshark shows that to be true right now running to your Plainfield, NJ
speedtest site, and my network peers directly with Comcast.
I'm kind of curious too. What is the purpose of this? Is it the
traditional purpose of CS1 to be less than best effort or something
else? If this is the case it seems Comcast would be purposely putting
themselves at a disadvantage in speed tests when congestion is
involved... or is this possibly on purpose to make peering problems look
even worse during congestion?
-Vinny
