Marketplace story on FCC and Comcast {for sure djf}

[David Farber <dave () farber net>]

________________________________________
From: Richard Bennett [richard () bennett com]
Sent: Monday, July 14, 2008 8:15 PM
To: David Farber; ip
Subject: Re: [IP] Marketplace story on FCC and Comcast {for sure djf}

I certainly agree with Tony that networks need to be designed to that heavy users don't unduly impact light users, and 
so that bulk data apps don't unduly impact interactive ones. The history of networking technologies is rife with 
examples of creating mechanisms to do this, and fixing bugs that prevented it. Two of my favorite examples from the 
lore: IBM Token Ring has a feature where the station immediately downstream from a high-traffic system, such as a 
server or gateway, had priority over other stations, because most tokens were released by the high-traffic system, and 
whoever was next in line got first dibs on the token. And in the world of the old coaxial cable Ethernet, we found that 
some vendors would boost their system priority by failing to observe the full 96 us inter-frame gap before jumping on 
the wire.

Problems like this arise in distributed systems in part because there's a huge element of trust in the overall design 
of a distributed system, many ways to violate it, and no mechanism for enforcement of equitable sharing. So we created 
a second generation of LAN systems that concentrated bandwidth allocation in switches that could enforce whatever 
policy the administrator wanted, which was typically nothing more than equitable sharing.

The problem that we have with the current debate on network neutrality is that many of the partisans are unaware of the 
fact that the Internet protocol stack does not contain any mechanism to ensure equitable sharing of bandwidth. In fact, 
one the foundation principles of the net neutrality movement is the concept of the "stupid network" that leaves all 
bandwidth management decisions to the endpoints and exhorts carriers to simply forward traffic first-come, first-served.

We engineers know that it's not that simple, and we know that simple FIFO policies in network routers are an invitation 
for abuse. The real solutions to NN are about defining parameters of enforceable fairness and stream classification 
policies, and communicating them to the lay public in a way they can understand.

The Internet protocols have always looked to me like an abstraction of Early Ethernet, before we had hubs, switches, 
VLANs, and priority traffic streams. There are good reasons to make a layer 3-7 stack look like this, but it's a huge 
mistake to pretend that layer 1 and 2 networks can operate with good stability and responsiveness with no more control 
than coaxial Ethernet had. We abandoned CSMA/CD for a good reason, and we can't have regulators forcing it back on us.

So NN is based on an engineering error, and we have to correct the NN misunderstanding of networking, the sooner the 
better. This doesn't mean that NN advocates are scurrilous charlatans, of course (although some are, they're a 
handful), just that we need more peace, love, and understanding.

RB

David Farber wrote:

________________________________________
From: Tony Lauck [tlauck () madriver com<mailto:tlauck () madriver com>]
Sent: Monday, July 14, 2008 11:35 AM
To: David Farber
Subject: Re: [IP] Marketplace story on FCC and Comcast

Those of us who make, or have made, our living in networking are
fortunate that appetite has been unlimited.  I see no reason for this to
change so long as computers, storage and communications remain on
similar technology curves.  If it isn't P2P, it will be some other
application, for example off-site backup. (If moving bits over the net
were cheaper than driving hard drives across town, I would be moving
tens of gigabytes regularly.)

Networks need to be designed so that the load placed by heavy users
doesn't unduly impact the responsiveness seen by light users. This is an
engineering problem that has been recognized since the design of
time-sharing scheduling algorithms in the 1960's.

Of course it would be easy enough to remove communications from the
technology curve, making the situation worse.  Two ways come to mind:
monopoly and regulation. It would be a shame to get one as the result of
trying to avoid the other.


Tony Lauck
www.aglauck.com<http://www.aglauck.com>




David Farber wrote:


________________________________________
From: Dr. Lawrence Roberts [lroberts () anagran com<mailto:lroberts () anagran com>]
Sent: Sunday, July 13, 2008 9:53 PM
To: Dan Lynch; David Farber; lroberts () anagran com<mailto:lroberts () anagran com>
Subject: Re: [IP] Marketplace story on FCC and Comcast

Dan,
For those who want to watch a movie per day, downloading a 3 GB movie at 500 Kbps takes 13 hrs, about all you want to 
spend on it. However you also must in return send out 3 GB of your stored movies each day for the system to work. If 5% 
of a University student body does this, a 20,000 student University would require 500 Mbps each way during the 13 peak 
hours. So you see, it is easy to consume 1/2 Mbps per person. Is there a limit? Yes I would think so. 2-3 movies per 
day seem like the most one could consume. But some of these students use 5 Mbps continuously. So when I say unlimited 
appetite, I mean within the context of current average people usage which is much lower.

An Example:
If a University has 5000 students and 180 of them use 1/2 Mbps for the peak period then the average data rate over a 
100 Mbps Internet access will be 20 Kbps and each P2P user is using 25 flows. There will be another 500 average users 
trying to operate one flow at any moment and they also will get 20 Kbps. 90% of the capacity is going to P2P.
Larry


At 03:56 PM 7/13/2008, Dan Lynch wrote:
Larry, I don't doubt your measurements, but gee, are you saying that P2P has
an "unlimited" appetite?  As if there are an unlimited number of files to be
shared and the queue stretches to infinity?  Where is that demand coming
from?  I am imagining hordes of files out there that people are adding to
their Torrent lists as the pipes come slightly unclogged. Is that your model
of demand?

Dan

PS.  Too bad storage got so cheap! Else someone could be making more money
out of that!


On 7/13/08 2:28 PM, "David Farber" <dave () farber net><mailto:dave () farber net> wrote:



________________________________________
From: Dr. Lawrence Roberts [lroberts () anagran com<mailto:lroberts () anagran com>]
Sent: Sunday, July 13, 2008 3:08 PM
To: David Farber; Dr. Lawrence Roberts
Subject: RE: Marketplace story on FCC and Comcast

Dave,
Sure! I want people to start understanding the dynamics of the problem.
Please add me also if that,s ok so I can see comments.
Larry


At 02:44 AM 7/13/2008, David Farber wrote:
Can I repost to my IP list?

________________________________________
From: Dr. Lawrence Roberts [lroberts () anagran com<mailto:lroberts () anagran com>]
Sent: Saturday, July 12, 2008 11:43 PM
Subject: RE: Marketplace story on FCC and Comcast

Jeff,
Like George, I agree the DSL and FTTX ISP's will not shift, but there is
another reason. The DSL companies have just as big or larger problem with P2P
since P2P expands to fill any capacity. In fact, as I have been testing and
modeling P2P I find it taking up even higher fractions of the capacity as the
total capacity expands. This is because each P2P app. can get more capacity
and it is designed to take all it can. In the Univerisities we have measured,
the P2P grows to between 95-98% of their Internet usage. It does this by
reducing the rate per flow lower and lower, which by virtue of the current
network design where all flows get equal capacity, drives the average rate per
flow for average users down to their rate. They then win by virtue of having
more flows, up to 1000 per user. I suspect they do not do this on cable since
the upstream capacity is only 10 Mbps and when it saturates, they must stop at
about 80%. But raise the capacity per user and the capacity of the upstream
choke point and watch out! P2P can consume virtually any capacity.
Larry




-------------------------------------------
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<http://www.listbox.com/><http://www.listbox.com/>




Tel. 707-967-0203   Cell  650-776-7313
My assistant is Dori Kirk   Tel. 707-255-7094  dori () lynch com<mailto:dori () lynch com>


Dr. Lawrence G. Roberts,  Ph:+1 650-906-8746,  W:  
www.anagran.com<http://www.anagran.com><http://www.anagran.com/><http://www.anagran.com/>, E:    lroberts () anagran 
com<mailto: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: 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




--
"Difficulties can never be greater than your capacity to solve them."
    - P. R. Sarkar

Anthony G. Lauck
PO Box 59
Warren, VT 05674
Southface 5 (for UPS and FedEX)
81 Park Ave
Warren, VT 05674
(802) 583-4405 (802) 329-2006 (FAX)




-------------------------------------------
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



--
Richard Bennett




-------------------------------------------
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