Net’s Numbers Man Surveys the Domains
The Internet is notoriously decentralized, but there is a handful of players who must be in alignment in order for it to work. One of them is Jon Postel, a computer scientist who heads the Internet Assigned Numbers Authority at USC’s Information Sciences Institute in Marina del Rey.
Postel joined the Internet’s unofficial board of governors in 1970, when he was a graduate student at UCLA working on what was then known as the Arpanet. Most of Postel’s work has occurred behind the scenes, but the recent controversy over how the Internet should evolve from an academic to a commercial network has put him in the limelight. When he conducted an experiment to test the Internet’s internal architecture earlier this month, his critics accused him of trying to hijack the global computer network.
Postel says he looks forward to getting back to his research on high-speed computer networks, which has been on the back burner while he and others debate the future of the Internet. In his first extensive interview, Postel told The Cutting Edge why the Internet’s inner workings have become such a hot topic.
Q: What is the Internet Assigned Numbers Authority?
A: The Internet Assigned Numbers Authority, or IANA, coordinates the computer protocols that need to be arranged or allocated uniquely so that different applications don’t collide with each other. The most visible example is domain names, like isi.edu or latimes.com.
*
Q: How does the system work?
A: At the very top level are the 13 root servers, which all have the same information about how to find domain names. The data is updated at Root Server A, which happens to be at Network Solutions [the Herndon, Va.-based company that administers the Internet domain name system under contract from the National Science Foundation]. Then all of the other root servers get their data from there.
*
Q: The future of Internet governance has become a very political issue recently. Are these just normal growing pains?
A: The Internet had its origins as a military research project and then as an academic activity, and now it’s overwhelmingly a commercial activity. It’s a substantial change in philosophy, so there are lots of reasons for people to participate in the discussion.
At first, it was OK for the National Science Foundation to hire one company to run the domain name system, and, since all of the users were at universities, the NSF just paid for it. Then a few years ago it turned out that 90% of the domains were being registered to companies, so we questioned why university research funds were paying for that. So there was a decision to charge fees, but then it started to look like a monopoly, so we said we should have competition. All of the discussion since then has been about what’s the most appropriate way to introduce competition.
*
Q: What solution did the Internet community propose?
A: A committee came up with a way to have a shared registry system so that many different domain name registrars could all use the same system and not have two different people registering for the same domain name. They also did a lot of work with trademark attorneys and came up with a way for dealing with trademark disputes in an international context.
Now there are 88 people who have signed up to be registrars, and they have all joined this organization called Core, the Council of Registrars. They’ve been off developing their software for the shared registry system. They’re pretty close to being ready to go.
*
Q: Can they go forward?
A: That’s another question. The U.S. government has decided they’re going to help, and Ira Magaziner was put in charge. At the end of January, this group came out with a proposal for what should happen, which is called the Green Paper.
*
Q: What is the main difference between the Green Paper and the Core proposal?
A: The orientation of the Green Paper is to find places to introduce competition. Core envisions one shared registry that would be used by all of the registrars, but the Green Paper indicates there could be several of them.
*
Q: So who is going to decide whether to use the proposal from Core or the proposals in the Green Paper?
A: The only way this can work is if a plan evolves that essentially everyone agrees to.
*
Q: Who absolutely has to be on board in order to implement a plan?
A: The key technical piece is that the root zone has to be consistent. But the root is split up among these 13 servers. Some are run in universities, and I’m pretty sure they would do whatever IANA said. Some are operated by government agencies, and they might be inclined to do whatever the government said. And then there are some operated by Network Solutions, and who knows where they come down?
*
Q: So will the decision be made by just a small group of people?
A: Probably not. But it’s hard to know who represents the end user in this game. There are 60 million or so Internet users, and who represents them?
*
Q: How soon will a decision be made?
A: It’s got to be within a couple of months, otherwise it will be difficult to hold the Core group together.
*
Q: With all of this going on, you did something a few weeks ago . . .
A: The experiment.
*
Q: Right. What were you testing?
A: IANA has to be prepared to run Root Server A because when Network Solutions becomes one of many competing registrars, it won’t be appropriate for them to be operating this essential resource. So we put up this machine that could be the primary root server and we wanted to verify that it would work. We asked some of the secondaries to get their data from this new IANA machine. After about a week we told them to switch back.
A lot of people from the government said, “How can IANA go and do this without asking me?” Well, IANA never asked them before about anything, so why would it occur to us to go and ask them this time?
*
Q: What do you think the chances are that a consensus can be reached about where to go from here?
A: I don’t think the chances are 100%, but I think it’s at least 50-50.
*
Q: What happens if there is no consensus?
A: I have no idea what will happen. It will be a big mess.
