-
Website
http://techliberation.com/ -
Original page
http://techliberation.com/2008/02/21/network-neutrality-end-to-end-principle/ -
Subscribe
All Comments -
Community
-
Top Commenters
-
MikeRT
195 comments · 6 points
-
eee_eff
803 comments · 8 points
-
mwendy
97 comments · 4 points
-
Ryan Radia
184 comments · 5 points
-
Richard Bennett
612 comments · 1 points
-
-
Popular Threads
-
Google on “Open”: Myopic Self-Focus
2 days ago · 7 comments
-
The Deontological Case Against Net Neutrality Regs
4 days ago · 16 comments
-
Facebook Privacy Controls Change & EPIC’s FTC Complaint
1 week ago · 10 comments
-
Cutting the Video Cord: “Apple TV” 2.0 + Disney & CBS
3 days ago · 3 comments
-
Google’s “Open” Philosophy and the Conspicuous Lack of Open-Source Search
1 day ago · 1 comment
-
Google on “Open”: Myopic Self-Focus
But there are other interpretations; Wu himself has defined it at least five different ways, one of them being "a principle of network architecture."
I don't think the DNS thing clarifies because we can't agree whether DNS is an end-point service or a network feature.
One thing that is clear is that the Internet is not really a pure end-to-end network, because it does routing in the core rather than at the end points (end point routing is called "source routing".) The PSTN is actually more end-to-end than the Internet because it lets you select your long distance carrier - route - at the end point.
The Internet does everything at the edge except when it doesn't.
Most network providers disallow source routing. Period.
To me network neutrality means simply that the purchaser of services at some level in the protocol traffic is entitled to have his traffic treated exactly like every other purchaser of the same service. If somebody wants to introduce a notion of "class of service" I guess that is OK, but the network is going to need some re-engineering because the way it works now, when anybody hands the traffic off, they pretty much lose control of it.
The issue is if you select 100 people using the Internet to change the DNS servers they use, most will have no idea what you are talking about.
Of the remainder, most will use (be required to use) DHCP or a cable-clone of it where the configuration received will carry the DNS server to use.
Of the remainder, most will not know an alternative to use. And a lot od fire-walls won't allow it.
And in any case, I don't think DNS has much to do with the neutrality issues.
What does have to do with it is policy routing where the "policies" involve source, destination, protocol, and things like that.
See "packet shaping" http://www.packeteer.com/
Can you tell me what these mean and what the distinction is?
Selecting your own DNS server is not incompatible with the use of DHCP. DHCP does provide a means for an ISP or network operator to propose DNS servers that users can use, and most DHCP clients will, by default, accept these suggestions and configure the local resolver to obey these suggestions. However, there is no reason that DHCP clients must do so, and they could be configured not to. For example, I was able to find out quickly by looking at the documentation for pump that pump.conf files can use the nodns directive to simply ignore DNS server information provided by DHCP servers (so that manually-configured DNS settings would continue to be used, even as IP address and route and network configuration changed). I'm sure that many other DHCP clients have an equivalent option.
For example, the OpenDNS people have configuration suggestions for various operating systems and they show that Windows XP separates the concepts of obtaining an IP address automatically and obtaining DNS server addresses automatically:
https://www.opendns.com/start?device=windows-xp
I imagine that the other instructions at
https://www.opendns.com/start
would provide fairly straightforward means of choosing to use OpenDNS in lieu of ISP-provided resolvers from many different operating systems, even while continuing to use DHCP.
(I'm not sure that this resolves the question of what kind of service we ought to consider the DNS to be.)
Free.net, in its original implementation, did violate the end to end principle, in that much processes (cloaking really) was done on the intermediate points.
I understand the interest in technology, but the heart of the network neutrality debate lies in the content and protocol neutrality issues: whether AT&T; will be able to stop criticisms of it by denying or delaying access to content they don't like.
Protocol neutrality is in fact very closely linked to content neutrality, more closely than many seem to think. A recent example is the Bank Julius Baer suppression of the site wikileaks. Of course I immediately went over to PirateBay and downloaded the torrent and helped redistribute the content the bank objected to, and have done the same thing for the training video that the RIAA has distributed to prosecutors.
If ISP's can discriminate by content or by protocol, I will lose the freedom to redistribute this type of content.
Of course, I can't see the totality of your paper, but I feel overfocusing on the technology and not on the results could result in your paper being obscurantist.
And there is considerable pressure to block port 53 traffic from end-user locations.
the e2ep specifies that the endpoints should be 'involved' in a process, but it is generally understood to exclude performance optimization, which QoS is an example of.
Even without the exception, apps would not be 'broken' by any selective QoS, and thus the e2e principle is not violated.