[governance] Regarding IPv6

[John Curran]

On Feb 6, 2011, at 12:51 AM, Avri Doria wrote:

Apologies for the belated reply... this list has a very respectable 
volume and my duties prevent me from always keeping up in real-time.
I'll respond to your post on IPv6 just for clarity, and perhaps come
back to merits of the IETF system as a governance model separately. 

> I have not researched it, just lived through it, especially in the years of the IETF.  I have been following, and participating at times, from the time the IPng issue got started.

Myself as well.  

Disclosure: I served as an IETF Area Director for the Operations and 
Network Management area and served as one of the 15 members of the 
IETF IPng Directorate.

> I have not written anything on it, though it may be an idea to do so someday, though there are so many things I would like to write someday when I not longer have to spend most of my time writing governance feuilletons for a living. Not likely, but a nice thought - I have so many things I never finished writing.

I think we all have a long list of future projects like this (although 
I do encourage you to move this higher up on your list... :-)

> Some of the chapters might include:
> 
> - the story starts with requirement  that were ignored when the beauty contest picked the winner.  and the fact that the winner was picked before it was time and the work was complete on the candidates due to political pressure to have a solution now. After in 1994 all the best people already knew that all IPv4 was going to die in a year or so.

I agree with the majority of the above (IPng was decided before all of the
important development work was completed, and hence some of the requirements,
e.g. transition, were left as future work), but I would take exception with 
the "all the best people already knew that all IPv4 was going to die in a 
year or so."   While IPng development was initiated as a result the various 
IPv4 runout analyses, I believe that it was well-understood most of those 
involved that fully developing a new version of the Internet protocol was 
going to take years; i.e. we were worried about dates and overall runout 
events out that were forecast out beyond 2010.

It is true at that time that we also knew that there was an imminent runout 
of class-B address blocks (potentially within 1 year) but that was going to 
be addressed with the concurrent activity of quickly deploying support for 
Classless InterDomain Routing (CIDR) across the Internet, i.e. I do not of 
anyone who thought of the potential runout of those blocks was any form of 
motivation for making an IPng decision "quickly".  I'll also note that at 
the time, there were complaints that the decision process seemed quite slow 
i.e. IETF direction on various network protocol documents usually takes 
two or three meetings, i.e. one year, and yet by 1994 work on various new 
version of IP had been going on for over two years without convergence.

> - It includes the inability of anyone to get people to commit to fixing the routing architecutre while going through the pain of changing the address structure.  This even though there were candidate solution that included routing architecture considerations.  Many of us believe that architecture and addressing must always be worked on together.

Agreed.

> - it includes a very strange tale of the inability of some very smart people to persuade anyone to include the notion of variable size addresses, or at least fixed addresses that allowed for IPv4 encapsulation.

Also agreed.  Please find me at a bar someday regarding this topic; I will
spare this list my rantings.

> - It goes through at least a decade of hubris where IPv6 was going to replace IPv4 any day now and the elite of IPv6 drank very expensive scotch to toast to the universal deployment of IPv6  (the fact that they drank Scotch instead of Irish was already a good clue that something was very wrong)

Anyone who thought IPv6 was going to deploy itself prior to the actual
IPv4 full depletion was quite confused, as we included no features to
distinguish IPv6 over IPv4 and thus create inherent motivation to deploy.
In 1994, I strongly noted this concern in my requirements paper to the
IPng Directorate (i.e. RFC 1669).

> - It includes years of miscalculation that IPv4 was the walking dead so there was no reason to think about coexistence.  

Mostly agreed; I'd have phrased it slightly differently, i.e. "years 
of miscalculation that the installed base of IPv4 plus 'imminent' IPv6 
deployment precluded updating IPv4 to allow for incremental deployment 
and coexistence."

> - it would include chapters of how CIDR and NAT saved IPv4.

CIDR and NAT only delay the IPv4's terminal sentence... there are still 
serious problems with trying to make IPv4 continue to work on the scale 
of the Internet today, and they are about to get much, much worse.

> - it would discuss the economics of IP addresses and the fact that even though the need to have strict hierarchy is no longer that great, a free market in IP addresses is still prohibited.  

I could write an entire book on this topic, but the short answer is that 
there's nothing preventing a free market in IP addresses except the ISP
community itself, we have a very successful limited market today, and we 
seem to have increasing movement towards more relaxed market constraints.  

> - it would discuss the sacred cow legacy IPv4 address blocks and the multicast blocks that have never really been exploited.

Yes, another case of miscalculation that "that the installed base of IPv4 
plus 'imminent' IPv6 deployment precluded updating IPv4 to allow maximize 
usability of the entire IPv4 space.

> - it would discuss the new reality where we need to support 2 protocol stacks and the routing infrastructure to support those 2 protocols.  We no longer have one Internet, we have 2 Internets that exist side by side, but are separate universes.  Good for router hardware sales, but not really an optimal solution.

Agreed (and it's actually worse than that, since there's also the costs and
overhead resulting from the addition of various broken translation solutions 
between the two.)

> - it would discuss that in the future, while there will be IPv6 in the network, there will be IPv4 for most of our lifetimes.  The routing architecture is still a disaster, and there will be many new solutions to keep IPv4 going, so that unless vendors switch over based on political pressure, many of us will continue to use IPv4 for a very long time to come.

IPv4 will be with us for a long-time, just as there are indeed folks still 
running SNA and X.25 even today.  As long as new customers can be connected 
up via IPv6 and have a functional Internet, I'll claim that IPv6 still meets
an important goal.  We are now seeing very serious engineering efforts underway
by carriers (both residential and mobile) to connect customers to the Internet
via IPv6, as is expected now that IPv4 depletion is actually upon us.

> As I said, IPv6 has come far enough so it will probably survive - I used to believe that could/should never happen but I long ago gave up fighting it.  But as we celebrate its alleged ascendancy, I just wanted to point out the sad path we took to get to this point and warn that we should not expect IPv6 to be the last address solution, nor should we expect that it will be an easy road ahead.

The road is not easy, but IPv4's 4.3 billion addresses doesn't adequately meet
the needs of 7 Billion people, as each becomes more Internet connected at home,
at work, and while mobile...  The enormous growth in online services, including
games, search, chat, publishing, etc. also results in vast server farms which 
require unique addressing.

> Me, I am still trying to figure out how to route on names and to avoid bothering with psuedo-numbers (aka IP addresses) anyway - why translate from one name type to another? Given that routing is getting less hierarchical all the time, these numerical names may someday be an anachronism of the past.  But this is a governance list and not technical research speculation list, so that was probably off topic.  

:-)

> And I apologize for being an IPv6 heretic and a party pooper.


Avri - I would call it being "an IPv6 realist"...  I am not very far apart in 
my perspective, but do see (finally) that IPv6 is actively being sought as a 
solution by service providers at this point in time, as the alternative approach
to adding customers via carrier grade NAT and IPv4 kludges is far less appealing.

FYI,
/John

Disclaimer:  My personal views alone.  No global namespaces were harmed in the 
creation of this message.  Attempting to maintain all global state locally may 
result in excessive stress or fatigue, and should not without first consulting 
a physician.




-------------- next part --------------
____________________________________________________________
You received this message as a subscriber on the list:
     governance at lists.igcaucus.org
To be removed from the list, visit:
     http://www.igcaucus.org/unsubscribing

For all other list information and functions, see:
     http://lists.igcaucus.org/info/governance
To edit your profile and to find the IGC's charter, see:
     http://www.igcaucus.org/

Translate this email: http://translate.google.com/translate_t