Today, the IETF Turns 25

FranckMartin writes "Little known to the general public, the Internet Engineering Task Force celebrates its 25th birthday on the 16th of January. DNSSEC, IDN, SIP, IPv6, HTTP, MPLS ... all acronyms that were codified at the IETF. But little known, one can argue the IETF does not exist; it just happens that people meet 3 times a year in some hotel around the world and are on mailing lists in between. The openness of the IETF and its structure has inspired the way ICANN is run, as well as the way the Internet Governance Forum (IGF) has been open to the civil society."

Comments

ICANN is open?

[QuietLagoon]

That's news to me.

Re:ICANN is open?

[dpilot]

Wouldn't things be much better and open if instead of ICANN or IETF we just had joint agreements between Comcast, ATT, Verizon, Time-Warner, and maybe Level3? I guess I should be international and include a few other "key players" in that list, too.

Re:ICANN is open?

[QuietLagoon]

Wouldn't things be much better and open if instead of ICANN or IETF

I'm OK with the IETF.

It is ICANN that has messed things up with their awful processes and back-room dealings.

Re:ICANN is open?

[Simon80]

http://meetings.icann.org/remote-participation

Re:ICANN is open?

[Tacvek]

You should not have any problem with the IETF. They do virtually everything in the public. You may read the mailing lists, and contribute.

The standards created by the IETF are generally completely open by every definition, in that any part may participate in development, there is no restrictions of any kind on access to the standards text[1], and most do not require any patent licensing[2].

The IETF has also been generally fairly competent. They did screw up a bit with IPv6[3], but otherwise many of the standards have been widely deployed with little or no issue, which is about all a standards body can ask for.

ICANN on the other hand has real issues. They are largely a policy organization, with the technical aspects restricted to the part known as the IANA. ICANN's IANA implementation leaves no complaints, but it does extremely poorly with setting policy. It does not help that it gets money from the US Government, which remains under the delusion that it is in charge of the internet. Personally I would kill off ICANN, having the IANA move under the umbrella of the ISOC, and have the IANA begin publishing the root zone directly, rather than involve Verisign and the US Government. Of course that would mean replacing the DNSSec key and procedures but that can be done.

Then we need to replace the policy aspect of ICANN, which is trickier. Policy should not be dictated by what would make the most money for the Registries and Registrars, assuming that the new policy would even keep that broken system. Better would be to require all Registries to be Non-profit organizations, which also operate as the sole registrar for said Registry. The exception would be the CC-TLDs, which would be run however the country in question desires.

Of course, the key here would be to convince the root server operators to go along with the coup, since ICANN would never relinquish control. Another tricky part would be for the new Registries to get the information needed to run (what domains are owned, and by who) from the existing for-profit registries.

Footnotes:

[1] One notable restriction it does not have is any fee to obtain a copy.
[2] Although the standards are unfortunately permitted to be developed to require known patents that will require a license with either upfront costs or royalties.
[3] They screwed up by not initially specifying some mechanism for IPv6 only clients to talk to IPv4 only hosts. This has resulted in multiple proposed mechanisms for doing just that, which are incompatible, and will cause some headache before we just go with one. They all include some sort of Network Protocol Translation (which obviously includes translating network addresses) being done by some ISP server or router. A few abuse other systems like DNS to do this, while the better ones merely encode IPv4 addresses in IPv6 addresses and use anycast to route them to the ISP servers/routers performing Network Protocol Translation.

They also screwed up by not taking the opportunity to fix anycast to make it support TCP connections. Anycast addresses are permitted as destination addresses, but source addressed must be unique to the device in question, so the TCP SYN/ACK packet would include the unique address of the server. If it were mandated that the TCP stack read this address and use it as the destionation IP for all remaining packets in this connection, then TCP would simply just work over IPv6 Anycast.

Re:ICANN is open?

[FranckMartin]

Wouldn't things be much better and open if instead of ICANN or IETF we just had joint agreements between Comcast, ATT, Verizon, Time-Warner, and maybe Level3? I guess I should be international and include a few other "key players" in that list, too.

It is called the ITU!

Re:ICANN is open?

[maXXwell]

ICANN has a history of antipathy toward public participation.

Kieren McCarthy, who has alternately been a journalist covering ICANN and also worked as ICANN's , general manager of public participation, has very good commentary on ICANNs merits (there are a few) and foibles (there are many):

http://kierenmccarthy.com/category/internet-governance/icann/

Re:ICANN is open?

[Anonymous Coward]

One of the characteristics of any open standards group is that people come and go at will, and multi-year projects (as IPng was) can have terrible times making progress because each new participant wants to re-start old conversations. That's dealt with in the IETF by a cadre of old-timers who say "read the archives" in loud tones until the new participant either reads the archives or goes away. It's a social defense to a potential DoS attack.

But it can fail horribly, and it did with IPng. The current design is a fork, as IPv4 addresses are not valid IPv6 addresses. Variable length addresses would have avoided that. They were considered early on in the design process and rejected because the silicon to handle it at core network speeds was expensive or impossible to make. To give you a sense of the network at the time, the number of routers which would have needed core network speeds at this decision point was on the order of *fifty*.

The design choice to use a new, longer, fixed-length address having been made, all sorts of other second system issues crept in, as participants saw it as the last chance to get something into the core internetwork protocol. Most, like mandatory IPSec, were later dropped, but the years dragged on nevertheless. Now, the silicon to do variable length address could be done easily at commodity prices. But we're stuck with a system with little to no network effect trying to start up against a system with incredible gravitational pull. So everyone sees 20 years of dual stack future, with that v4 address space living on and the v6 space being assigned and reachability built up over decades. It was a flag day design, in a network with no chance of a flag day like the NCP transition.

Needless to say, lots of people pointed out the different choice as things moved through the decade plus, but the cadre (especially the chairs of the v6 working group) tolerated no reconsideration. LiSP is being built elsewhere in the IETF and may create an overlay whose reachability includes both sides of the dual stack, but it is very late.

The IETF works reasonably well, but it is slow and reconsiders its decisions poorly, for reasons common to "rough consensus" organizations. And sometimes the price paid for that is awfully high.

Oh, the pain

[whoda]

My brain melted trying to comprehend that summary on a Sunday morning.

Re:Oh, the pain

[Gripp]

lol - try reading one of their RFC's!

Re:Oh, the pain

[Anonymous Coward]

lol - try reading one of their RFC's!

try writing one! The XML format for authoring an RFC is ridiculous!

Re:Oh, the pain

[DeathSquid]

Having authored several internet drafts, I found the XML format perfectly reasonable.

The IETF is engineering for the long term future. If the RFCs were written in some proprietary format, many would be effectively unreadable today. Imagine if they had decided to standardize on Framemaker or Wordperfect 25 years ago. Instead, the decision to use plain ASCII text for the corpus of RFCs means that we will be able to read these documents for the foreseeable future.

ASCII text isn't perfect. Tables and diagrams are awkward and you can forget about international characters (although UTF-8 seems like a reasonable future upgrade). But the benefits far outweigh the impositions

Interestingly, you are still only required to submit internet drafts as plain ASCII text. The XML tool you complain about simply makes it easy to conform to the prescribed formatting conventions and structure. So please feel free to ignore it.

Re:Oh, the pain

[Anonymous Coward]

I think what I am really complaining about isn't the format of the document as much as the interface to authoring it. It would be helpful if there were an emacs-mode for it, or some helper tools instead of being required to do it in a plain document.

Huh?

[Frosty Piss]

ZzzzZZzzz... KASNORK!!! ZZZZzzzzzZZZZZZzzzz...

I'm sorry, what? Who? ...OK...

Why did you wake me up?

IETF doesn't exist

[Anonymous Coward]

IETF is the shadow-ICANN. The masters behind the veil...

Anyway, congrats IETF. May your next 25-years be even better than the first 25 - when the said standards listed in the summary are actually implemented as widely as HTTP.

Re:IETF doesn't exist

[Tacvek]

The IETF exists in the same way that Debian Project exists, as an unregistered association of individuals, who operate under specific rules created by consensus. However, not being a legal entity in and of themselves, they each have umbrella organizations, namely Software in the Public Interest and the Internet Society, respectively. Both umbrella organizations encompass multiple other proects in addition to Debian and IETF.

The only notable difference is that while the Debian Project has clearly defined members (the Debian Developers), the IETF does not. On the other hand Debian relies heavily on individuals who are not members, making even this distinction smaller than it may seem.

Re:IETF doesn't exist

[mbone]

The IETF does not legally exist, and has no members (legal or otherwise). The IETF Trust does legally exit, to hold IETF copyrights and trademarks, and ISOC (a 501(3)c charity) serves to support the IETF, although there is nothing legally binding the IETF to ISOC (nor could there be, as non-existant entities cannot enter into binding contracts).

Re:IETF doesn't exist

[Tacvek]

Which is very similar to the Debian Project, since it also does not legally exist. While it does have membership as I pointed out, it does not really mean that much since it is highly reliant on non-members. (The IETF is obviously entirely reliant on non-members since it has no members).

SPI (Software in the Public Interest) serves both roles (to hold the trademarks and other resources including funds, as well as to provide support) for Debian, which likewise has no binding legal ties to SPI, and could not have such ties for the very same reason.

Thus I'm quite sure we are both saying the same thing, you saying it directly, and me by analogy to another non-(legal-entity) that some people may be more familiar with than the IETF.

Re:IETF doesn't exist

[mbone]

Or, to put it another way, the IETF does not exist in the same way that the Debian Project does not exist.

There are evidently classes of non-existance, as the IETF does not not exist in the same way that (say) Rivendell does not exist.

This isn't the IETF I've heard about

[Anonymous Coward]

I've heard a very different version of this story

Re:This isn't the IETF I've heard about

[Anonymous Coward]

This is the Star Trek version - you know, the altruistic peace and harmony and in the future we use no money and everything is for the benefit of mankind - that version of the IETF

Re:This isn't the IETF I've heard about

[mbone]

Well, there is the International Essential Tremor Foundation.

That is a rather different IETF.

ICANT

[girlintraining]

The openness of the IETF and its structure has inspired the way ICANN is run,

Yes, I believe for the ICANN people, it served as a giant lighthouse warning petty tyrants of the dangers of open, collaborative design processes. Since ICANN took office, domain name registration has become horribly convoluted, the prices have gone up, lawsuits abound, and we're now in danger of running out of real estate (IPv4 addresses), while they sit on their arse and worry about copyright. They're like a HOA -- they're fining people left and right and ordering them to take down christmas decorations, flags, and people who dare to paint their house in an unapproved color, while they forget to spend money on things like garbage collection, road repair, and snow removal.

No, actually, they ARE the internet's HOA, and about as bloody useful.

Re:ICANT

[dodobh]

What does ICANN have to do with IPv4 running out? IP is IETF territory, and the IETF has done it's job by putting IPv6 out there.

IPv6 adoption is now a business problem, and you can see the lack of business innovation in that space.

all proposals must have working prototypes before?

[petithug]

This sentence in the article is no longer true. There was a time when people coded stuff, then wrote an I-D to document it. The problem is that the burden of having at least 2 implementations is only to promote an RFC to the Draft Standard level, which is less and less frequent.

This is a real problem, because some of the bugs in an RFC can only be found by testing two implementations against each other. Unfortunately my last tentative to improve this was rejected:

http://www.ietf.org/mail-archive/web/ietf/current/msg55964.html

WTF

[Anonymous Coward]

"... MPLS ... all acronyms that were codified at the IETF"

Huh? The IETF invented the Twin Cities? (Minneapolis-St.Paul)