Re: do read !!! Comcast confirms 250GB cap effective October 1

[David Farber <dave () farber net>]

Begin forwarded message:

From: Tom Vest <tvest () eyeconomics com>
Date: September 2, 2008 4:07:19 AM EDT
To: dave () farber net
Cc: "ip" <ip () v2 listbox com>
Subject: Re: [IP] Re:    do read !!!  Comcast confirms 250GB cap  
effective October 1

Hi Dave,

Extremely credible local sources in both Japan and Korea express grave  
doubts about the validity of this explanation, at least for those two  
markets at the current moment in time. In both cases, the very high  
quality/high density of local infrastructure, coupled with the very  
high correlation of language and content demand are interacting with  
p2p algorithms as one would expect, to produce the same kind of highly  
localized geographic distribution patterns that one generally observes  
with more conventional content distribution systems in those markets  
-- or so the local experts suggest.

International traffic flows are apparently growing faster than local- 
local traffic exchange, at least in Japan, because of an atypical  
spike in the popularity of specific "foreign" content sources -- e.g.,  
Google Maps, Youtube, et al. -- one that has (perhaps only  
temporarily) outpaced the natural tendency for foreign operators to  
localize such flows themselves.

Having gotten out of the localizing-traffic-in-East-Asia business  
myself a few years ago, I was skeptical but didn't trust my own  
doubts. If there are any other questions or comments on this thread,  
I'll see if I can persuade the current experts themselves to chime in.

If one wants to convince the world of something, perhaps the first  
place to start is with a convincing understanding of the world itself...

TV

On Aug 30, 2008, at 8:33 PM, David Farber wrote:

> Begin forwarded message:
>
> From: "Dr. Lawrence Roberts" <lroberts () anagran com>
> Date: August 30, 2008 1:16:14 AM EDT
> To: David Farber <dave () farber net>, Larry Roberts <lroberts () anagran com 
> >, dan () lynch com
> Subject: Re: Fwd: [IP] do read !!!  Comcast confirms 250GB cap  
> effective October 1
>
> Unfortunately, the problem is even worse for Japan and S. Korea. P2P  
> users are more common, the countries are small so most downloads or  
> uploads go international, and their High speed symmetric BB links  
> suffer far worse than our un-symmetric lower speed BB connections.  
> So they have major economic problems. They have tried DPI even  
> before us but it cannot find all the encrypted flows now so the  
> remaining 30% or the P2P users still can operate freely. The  
> remaining P2P users then spawn more flows and take up the same  
> 80-90% of the pooled capacity. The result is the average user get  
> very poor service.
>
> However going to a GB cap or a charge per GB is really bad as we  
> know. But put this down to flailing for a solution in the face of an  
> extremely bad situation. The Internet cannot stand up to this as  
> more and more applications discover the ease of getting more  
> capacity by using more flows.
>
> I continue to believe that equal capacity for equal pay instead of  
> equal capacity per flow is what we need to move to, since then  
> congestion will affect us all fairly (as it used to). But how does  
> one convince the world?
> Larry
>
> At 02:12 PM 8/29/2008, David Farber wrote:
>
>
> > Begin forwarded message:
> >
> > From: Dan Lynch <dan () lynch com>
> > Date: August 29, 2008 2:37:05 PM EDT
> > To: Dave Farber <dave () farber net>, Tony Lauck <tlauck () madriver com>
> > Subject: Re: [IP] do read !!!  Comcast confirms 250GB cap effective
> > October 1
> >
> > Thanks Tony.  Now I think it gets trickier here as a solution for a
> > single
> > AS (like Comcast) may not be very helpful for customers who span
> > multiple
> > Ases, if the intermediate systems do not also play by the same rules.
> > Maybe
> > someone like Mike O'Dell can offer his deep experience here?  Or  
> > Larry
> > Roberts whose new company, Anagran, is wrestling with this issue of
> > "fairness" across a long haul.  We do need some "rules of the road"  
> > that
> > benefit everyone or no one will win very much.  Right now those
> > countries
> > that are overprovisioning (like Japan, South Korea and Finland) look
> > pretty
> > smart and don't have to address the fairness issue (yet).
> >
> > Dan
> >
> >
> > On 8/29/08 10:13 AM, "Dave Farber" <dave () farber net> wrote:
> >
> > > Begin forwarded message:
> > >
> > > From: Tony Lauck <tlauck () madriver com>
> > > Date: August 29, 2008 12:55:28 PM EDT
> > > To: Dan Lynch <dan () lynch com>
> > > Cc: "'David Farber'" <dave () farber net>
> > > Subject: Re: [IP] Re:   do read Comcast confirms 250GB cap effective
> > > October 1
> > >
> > > Dan,
> > >
> > > Yes, a cap will solve the congestion problem if it is set  
> > > sufficiently
> > > low.  In the limit it definitely solves the congestion problem by
> > > driving away all the customers. (The Prodigy solution.)
> > >
> > > The problem is that congestion is caused by instantaneous load
> > > exceeding capacity, whereas monthly usage measures average load. If
> > > there are, say, four busy hours per day, an average based cap will
> > > need to be set (roughly) six times lower than it would otherwise  
> > > need
> > > to be. The result will be unhappy customers. In addition, the bean
> > > counters (who are unlikely to know any queuing theory) may suggest
> > > that expenses be cut by reducing resources!
> > >
> > > I have written many IP posts on this subject and you can find them  
> > > in
> > > the archive. Basically, a proper solution is to give higher priority
> > > to users who create less congestion. Let everybody go as fast as  
> > > they
> > > want when there is no congestion. If people need or want more  
> > > capacity
> > > than others, let them pay more for more "shares" and adjust the
> > > weighting accordingly, and use the money to increase capacity. This
> > > approach is politically neutral, i.e. it will work in a capitalist  
> > > or
> > > a socialist society. Unlike the Comcast cap it follows the principle
> > > of maximum utilization of resources.
> > >
> > > Tony Lauck
> > > https://www.aglauck.com
> > >
> > >
> > >
> > > Dan Lynch wrote:
> > > > Tony, can you teach me how a cap does not solve their congestion
> > > > problem?
> > > > In the limit it sure does...
> > > > Thanks,
> > > > Dan
> > >
> > >
> > >
> > >
> > > -------------------------------------------
> > > Archives: https://www.listbox.com/member/archive/247/=now
> > > RSS Feed: https://www.listbox.com/member/archive/rss/247/
> > > Powered by Listbox: http://www.listbox.com
> >
> >
> >
> > Tel. 707-967-0203   Cell  650-776-7313
> > My assistant is Dori Kirk   Tel. 707-255-7094  dori () lynch com
>
>
> Dr. Lawrence G. Roberts,  Ph:+1 650-906-8746,  W:  www.anagran.com,  
> E:    lroberts () anagran com
> Founder, Chairman, Chief Architect, Anagran, Inc., 580 Pastoria  
> Ave., Sunnyvale, CA 94085 USA
> If not for you, please return. Any use other than the intended  
> recipient is unauthorized.
>
>
>
> Archives





-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com