[173609] in North American Network Operators' Group
Re: Many players make up application performance (was Re: Richard
daemon@ATHENA.MIT.EDU (Paul WALL)
Tue Jul 29 14:20:08 2014
X-Original-To: nanog@nanog.org
In-Reply-To: <CFFD4F34.16874E%kevin_mcelearney@cable.comcast.com>
Date: Tue, 29 Jul 2014 18:20:01 +0000
From: Paul WALL <pauldotwall@gmail.com>
To: "McElearney, Kevin" <Kevin_McElearney@cable.comcast.com>
Cc: "nanog@nanog.org" <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org
The devil is in the details. Ken Florance
(http://blog.netflix.com/2014/04/the-case-against-isp-tolls.html)
paints a different picture in his blog, for example.
As a manager at Comcast, can you refer the people on this list to any
ISPs who do not have a history of congestion into your network? This
question comes up about once a month, absent any good solutions, so
insight would be appreciated.
Drive Slow,
Paul Wall
On Tue, Jul 29, 2014 at 5:25 PM, McElearney, Kevin
<Kevin_McElearney@cable.comcast.com> wrote:
>
>
> On 7/29/14, 12:45 PM, "Valdis.Kletnieks@vt.edu" <Valdis.Kletnieks@vt.edu>
> wrote:
>
>>On Tue, 29 Jul 2014 14:33:28 -0000, "McElearney, Kevin" said:
>>
>>> (w/ a level of quality). <$IP_PROVIDER> plays a big role in delivering
>>> your *overall* Internet experience, but eyecandysource plays an even
>>> bigger role delivering your *specific* eyecandy experience. If
>>> eyecandystore has internal challenges, business negotiation/policy
>>> objectives, or uses poor adaptive routing path decisions, this has a
>>> direct and material impact to your *specific* eyecandy experience (and
>>> some have found fixable by hiding your source IP with a VPN).
>>
>>Very true. But what we're discussing here is the *specific* case where
>>eyecandystore's biggest challenge at delivering the experience is an
>>external
>>challenge, namely that $IP_PROVIDER's service sucks. It's particularly
>>galling when $IP_PROVIDER's internal net is actually up to snuff, but the=
y
>>engage in shakedown tactics to upgrade peering points.
>
>
> There is a great analysis by Dr Clark (MIT) and CAIDA which shows while
> there are some challenged paths and relationships between providers, this
> is the exception vs the rule. Using the =E2=80=9Cexceptions" are busines=
s
> decisions.
>
> Performance is a two way street (as are shakedowns)
>
> - Kevin
>
