Log in

No account? Create an account

Previous Entry | Next Entry

The IPv6 debacle

I've been reading some of the mailing list archives, drafts, and RFCs for IPv6, and in general, I'm somewhat disappointed (although intrigued). Most of the arguments that are being raised are not much different than those that were raised 15 years ago, maybe more. The same people are making the same arguments. Not much progress has been made; there are some networks where one can get IPv6 connectivity, but it's mostly using tunnels which are slower and less robust than native connectivity. (In this use of tunnels, the IPv6 packet is carried as the "payload", ie. the data inside the ubiquitous IPv4 packet.) There is not much of a compelling reason for anyone to switch to IPv6. Most people can get NATs to sit between the public IPv4 Internet and their organization or home networks, if they do not have them already.

Here is an overview of what has gone wrong with the IPv6 migration thus far.

Yet I am intrigued by the discussions. Some of the people work for organizations that are principal players in the migration, such as ISPs, address registries, and router vendors. It is part of their job to participate in these discussions, but they also are free to express personal points of view if they wish (as long as they provide suitable disclaimers). I am somewhat envious of these people; I wish I could have a job where I was able to discuss architectural issues like this (although I tend to tire of long flamewars where nothing is resolved).

To contrast these debates with similar debates on click fraud: Most of the participants in the IPv6 debates have considerable knowledge and understanding of the technical, legal, and business ramifications of various outcomes. They are also comfortable with airing their opinions in public forums; in fact, openness is highly regarded within the community. In click fraud debates, on the other hand, one-dimensional business perspectives tend to dominate, such as a common web publisher's credo of "What happens after the user clicks on the ad is not my responsibility." There is almost no openness or transparency (especially from the engines and ad networks). To some extent, this is understandable; it would not be appropriate to discuss specific fraud detection techniques that could be circumvented. But it should be possible to discuss the relative effectiveness of general techniques, given the limitations of the Internet architecture.



( 2 comments — Leave a comment )
Oct. 5th, 2007 08:35 am (UTC)
You are wrong
D.J. Bernstein is too far from reality - he is no longer working on internet protocols:

1. Basic interoperability issues - answer use dual stack

2. Incompatibility - IPv6 is not incompatible at UDP/TCP/SCTP and application level - what is the mistake to use 32bit for select services - use names instead of 32 or 128 bit selector

3. incoherence - you cannot expect coherence in IPv6 if in IPv4 you already don't have it. Every player in the Internet is using and abusing IP addresses differently.. There no silver bullet that fits for everybody - some transition methodologies are fitting in a particular scenario. There is a freedom to develop others. Without experiment like 6NET, Euro6IX, Moonv6 this would not have been possible...

4. distractions - it will be sad, that several user can talk to a certain services via proxy only because of lack of address space... We still have time to make IPv6 universally deployable....
Oct. 6th, 2007 03:57 am (UTC)
Re: You are wrong
Regardless of whether DJB is now working on protocols, he raised some valid points that are still true today. Note also that he is describing a process that had already been ongoing for over eight years (assuming the Last-Mod date on the page is correct). He is hardly the only person complaining about the transition; if you search on [IPv6 debacle] you'll get similar, more recent comments from top Internet architects. (The link points to the start of the thread, not the message that includes the text [IPv6 debacle].)

As to your other points, the lack of mass adoption of IPv6 speaks for itself.
( 2 comments — Leave a comment )

Latest Month

December 2017

Page Summary

Powered by LiveJournal.com
Designed by Tiffany Chow