NNSquad - Network Neutrality Squad

NNSquad Home Page

NNSquad Mailing List Information

 


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ NNSquad ] Re: "discrimination based on user-history" - will fairness make congestion the norm?


Wes Felter wrote:
Even the latest proposals to move away from protocol-agnostic
management techniques to "fair share" management have been
criticized as discriminatory against particular users; there is
currently an active discussion about this between certain
participants in the IETF's P2P Infrastructure Workshop.  And
according to some advocates who participated in the P2Pi Workshop,
"discrimination based on user-history is no better than
protocol-discriminatory behavior.""

Here is the source, from Robb Topolski:

http://www.ietf.org/mail-archive/web/p2pi/current/msg00072.html

I think it's worth reading the entire thread.

This is one more reason why I am leaning more and more toward allowing ISPs (as opposed to backbone providers) a lot of freedom in traffic management. Not _total_ freedom: I think that the FCC should take action if/when an ISP's "traffic management" amounts to completely blocking some application, or heavily throttling some application even at low-traffic times (as has been reported wrt Comcast).


Penalizing applications because of some perceived problem ("all P2P users are copyright violators" or "our bandwidth problems are all caused by BitTorrent, so we'll randomly disrupt its connections regardless of whether or not we have available bandwidth") amounts to the kind of "punitive" contract clauses that are considered unenforceable in commercial law.

But when traffic really is heavy, I think that Comcast, TW, Cox, etc. should be able to
a) prioritize based on protocol/port
b) prioritize to favor less heavy users
c) prioritize to favor users who pay more
d) etc.
e) combinations of the above


If they piss off too many customers, they will pay the price in decreased sales. If they find the right mix, in terms of keeping most customers happy and encouraging the heaviest users to pay more for more service, they will make more money. This is how capitalism is supposed to work.

Note that this applies to "burst usage" sales of the sort provided by most cable companies. If an ISP offers "guaranteed bandwidth" (a la DSL, frex) then you should get that much bandwidth, and it should have purchased enough bandwidth from its upstream providers to ensure that bandwidth at all times.

Note this applies to ISPs, where the end-user has a direct contractual relationship. I think that backbone providers should still be subject to common carrier rules -- at least as long as they are selling bandwidth as opposed to more complex services.

Part of my reasoning is basic free-market principles. Part of it is analogy. If I have a home firewall installed, it can and does look in my packets and decide which ones should get forwarded. In principle, it could also prioritize them, although I don't think my current one does. The same principle applies when I use the Internet at work. My connection goes through my employer's firewall, which may apply its own rules as to what packets get through, and I don't even have the privilege of replacing it or changing the parameters -- somebody else decides what the rule are, and it is "bundled" with my employment. In principle, my employer could even terminate my employment if they don't like the way I use their Internet connection.

In the same way, I have a contractual relationship with my ISP, which I am free to terminate if I don't like the way they manage their network. So they should be free to (within reason) decide which packets have higher priority, etc.

This would allow combinations that IMHO would be beneficial. E.g., the kind of usage-based prioritization that Topolski objects to, combined with something that picks off VOIP connections(*) and gives _those_ packets a higher priority.

(*) This could be as simple as inspection for port number, or it could be a full-fledged proxy that will only carry "real" VOIP protocol and won't "work" if somebody decides to put BT traffic on a port nominally designated for VOIP. Or combined with a usage measurement that lets VOIP traffic go through at high priority, while throttling it if it exceeds typical VOIP usage by (say) a 2:1 margin.

*At the same time*, I note that many of the desired features can be obtained by the user-assigned priority scheme that others have suggested. _You_ can assign (say) priority levels 1-4 (1 being highest) to your traffic, either statically (VOIP = priority 1, HTTP = priority 2, BT = priority 4, or whatever) or by letting the application set the priority. But these priorities would apply only _relative_ to your own usage. If you are using more than your "fair share", your highest priority packets will go through first, but if you assign P1 to _all_ your traffic, it's the same as if it were all P4.

This would still let the poor woman with the huge upload make her VOIP call for a medical emergency (or just to chat with her cousin) even when throttling is being applied because the ISP's network is overloaded and she's a "heavy" user at the moment. As long as her system is reasonably configured. If her husband has set up his uploads to be P1, he probably gets what he deserves. (Possibly a candidate for a Darwin award?)