A Concise Guide to the Major Internet Bodies

alex simonelis submitted a good summary of the major internet bodies. If you hunger to know the difference between ICANN, IETF, ISOC and the rest of the alphabet soup of the governing bodies that make our beloved internet possible, this is a great place to look. It covers 10 major organizations.

Comments

What vs How

[fembots]

It's nice to know what each organistion does, but is there an article about how they actually do it?

For instance, how (pardon my ignorance) ICANN actually controls numbers and names, technically. Is there a mainframe of some sort that stores them? How does ICANN make changes?

Re:What vs How

[Zeinfeld]

At the very top are the 13 root servers, run by people like VeriSign. If you want to make the Internet pretty useless, take out those servers (someone tried a couple of years ago).

Someone tries every day, to be more precise there are over 1000 attacks against core DNS each day. Most of the roots are run on a basis that I regard as far too casual given the critical nature of the infrastructure.

There are not 13 root DNS servers, there are 13 root IP addresses which is not the same thing at all. Several of the roots are anycast so there are actually multiple data centers serving them. The number of root servers is much larger than 13.

Another pretty major omission from the list is OASIS which has roughly the same degree of influence as W3C and considerably more than the IETF.

The premise of the list is somewhat misguided. The standards bodies themselves don't have any influence on the Internet, its the members and the software providers who have influence. The point of the standards work is to get buy in from the necessary stakeholders, not to solve problems by committee.

Giving the choice between having my spec rejected by the IESG and having it rejected by Microsoft or the Apache group I'll choose the first. One of the big problems with the IETF is that many folk think that they are somehow 'in control'. Not on this Internet you ain't, if I don't get a chance to vote on who holds an office I don't see why I have to respect the decisions made by the office holder. I certainly don't see why I should wait two years or more for them to come to a decision.

I helped set up W3C when the IETF web standards effort collapsed. HTML was originally proposed in the IETF and turned into a disaster. When W3C was not interested in doing the work I do I played a leading role in one of the early OASIS Web Security standards. I am currently sitting in a W3C working group where the discussion has got into some particularly arcane details of XML.

Standards organizations are a vehicle, they are neither the driver, nor the road.

ISOC/IETF vs ICANN

[elfuq]

No one (and this article itself) has ever really objectively described the compromises/disputes between the old internet governance infrastructure and the increasingly corporate-dominated and somewhat authoritarian ICANN.

ICANN is supposed to have a standards pillar. However all internet standards are really developed by the IETF, published by the RFC Editor and adopted by the community the way that they have always been. (The exception being HTML/HTTP and its derivatives - the W3C is entirely corporate)

There's some mention here of the dispute over IANA. Back in the day, it was just Postel, and he demonstrated entire control over the root servers. But now it's really not clear who controls the root servers, allocates IP address ranges to the regional registries, and assigns other numbers. This stuff should be transparent!

Re:ISOC/IETF vs ICANN

[C10H14N2]

Postel may be gone, but IANA, ICANN and IAB still have the same address... as for transparency, there's a lovely little explanation of how-it-works here:

http://www.iana.org/root-management.htm ...as for anything else, erm, if you're really so concerned, have you ASKED?

http://www.iana.org/contact.htm

It seems most people love to bitch piss and moan about ICANN/IANA, but they can't pick up a damned phone or write an email (or, for that matter, type in the F@#$ing URL that is rather forthcoming about process, policies and procedures) when it's far easier to scream "conspiracy!" ...god only knows how many tinfoil hat looneys already ring the phone off the hook. Hell, IMHO, it's pretty superhuman of them to publish their address and front-office phone number in the first place...and godlike that they still get a damned thing done as a result.

Re:ISOC/IETF vs ICANN

[rs79]

"It seems most people love to bitch piss and moan about ICANN/IANA, but they can't pick up a damned phone or write an email"

If you'd actually tried this you know how non-productive this idea is. I wasted 10 years of my life doing exactly this, only to watch a bloated and corrupt ICANN emerge in spite of everything hundreds of people did and now watch all our predictions about their future potential wrongdoings come true. I do not feel good about this.

When the US government was handing over IANA to "the new corporation" (ICANN), open-rsc (ORSC) was invited to "advise" them in the early days. What this amounted to was every time ICANN had a truly horrific idea and we pointed out what a bad idea this was ICANN simply took our adive by using it against us and found ways to work around our criticism by adding enough spin to makie it appear it wasn't a problem and of course they were prepared now to hear this criticism and had stock nonsensical answers for their PR machine.

The list of reasons why ICANN is an utter and abject failure is miles long. Put as succinctly as I can, ICANN is supposed to measure consensus and enact policy based on it. Anybody who as at or saw the Marina del Rey 2000 selection of the 7 new tlds knows how far from reality this actually is.

The video of this is still, I believe, available at the Berkman center. Worth a look...

As for conspiracy, I can give you names of people who have first hand evidence of it. Names you will recognize. It runs in the tens (hundreds) of millions of dollars and is he reason there are so few, and so lame, new tlds.

ICANN was born "behind the scenes" and has always operated that way. It gets more distane from reality every day. At some point it will implode.

Keep in mind ICANN, a $50M+ (and climbing) organization replaces what Jon Postel used to do as a part time "task" for about 30K a year. Except Jon could not be bought. Pity he died right before ICANN was formed. For the record, I first called him in 1994.

Re:My persistant question.

[The Cisco Kid]

Note that none of the entities discussed in the article is a major network operator - while they certainly may have their own organizational network just like any other company or organization, they dont directly operate the backbone networks. Their roles are advisory and (sort of) regulatory. To avoid any sort of appearance of favoritisim, I doubt they even get any special deals from whatever ISP they use to host their sites or connect their offices.

There is no 'center' or 'trunk' of the Internet. Every bandwidth flow is between two endpoints. Large backbone network operators generally have peering agreements (eg I'll send traffic to you that wants to go to your addresses if you agree to do the same for me, and we'll do it over the same set of wires) and either in most cases any two organizations that consider themselves to be 'peers' figure that average traffic in both directions will be the same, so they do it on a basis of each network paying for its own costs to interconnect to the other. Sometimes if the traffic is expected to be unbalanced, there will be a cost recovery clause in the peering agreement.

There are facilities known as 'peering points' that manage and operate various sorts of switched networks (FDDI, ATM, etc) that an organization can colocate routing equipment, and then have a shared 'connection' that they are able to use to peer with any other network operators that are located there. These are known as 'NAPs' - some were established back in the days of the NSF, some came later. These are about as close to the 'center' of the Internet as you can get, but they are not the center (nor is there a free ride to anywhere else from them)

Note you have to have your *own* IP addresses to peer, you announce your networks via BGP and accept announcements from your peers - you are specifically NOT allowed to use any other peer's router as your 'default route' - you can only send traffic to them that has a destination of one of the networks they announce to you as theirs, and you generally can only become party to a peering agreement if the other parties think you really are their 'peer' eg that it is desirable for them to connect to you as it is to you to connect to them. This would generally be met by being a large backbone yourself, with your own connectivity it multiple (more than 3) peering points, and your own customers (such as ISP's, webhosts, businesses, etc)

It is also possible to connect and a peering point and obtain what you think of as 'Internet service' - its called 'transit' - and its another type of agreement you can enter, that specifically *does* allow you to 'default' to the router of the org that you pay for transit. You can expect to pay market rates for transit bandwidth, although its a pretty competitive market. You would still be responsible for locating your own router onsite, interconnections with the shared fabric, and then the backhaul to your location.

For an interesting read, see http://worldofends.com/