New head of Internet Architecture Board speaks out

The Internet Engineering Steering Group has responsibility for the standards and the day-to-day operation of the IETF, the working groups, and so on. The IAB has responsibility for taking a longer-term architectural view and worrying about strategic questions. The IAB chair's role is coordination, agenda setting and becoming the point person on a lot of external relationships.

How does your IAB role fit in with your position as Internet architecture vice president at AT&T?

The IAB position is volunteer work. AT&T has been kind, generous and patient enough to give me the time and resources to do it.

What issues do you think are most important for the IAB to consider?

Scaling, scaling and scaling. And I hope that's the same answer you would have gotten two or three years ago. The problem continues, and it's an inevitable consequence of Internet growth. It's scaling with regard to how we operate the IETF, as the demand for protocol development increases and the number of people who are able to work - both technically and from a resource standpoint - doesn't increase nearly as quickly.

In scaling the network, there are questions as to how much longer some of the fundamental routing protocols are going to work given the level of knowledge they require. The administrative issues and the ICANN [Internet Corporation for Assigned Names and Numbers] situation are, in some respects, merely another scaling problem. Arguably, internationalization is another kind of scaling issue. IPv6, of course, is a scaling issue. The IAB is trying to do everything we can to make certain that IPv6 and other solutions that prevent our running out of address space are getting fair consideration but not in a context of being a magic bullet.

As solutions evolve in some of these areas, we have to make certain that they're consistent with scaling the network rather than being solutions that fractionate. These internationalization issues are a typical example. It's vitally important that people be able to communicate on the Internet in their own languages and use facilities based in those languages. At the same time, it would be quite unfortunate if the potential for human communication and e-commerce between countries were suddenly frustrated because we set up language and naming barriers.

Another key role of the IAB is to look at emerging activities in the IETF. Which ones do you find most interesting?

I certainly see working with nonhard-wired devices and mobility and portability as very important. I see the continuing evolution in how we name and identify things on the Web and elsewhere as being an important problem that we are beginning to make some progress on. The Domain Name System security issues are very important. Internationalization is vital.

I am personally very worried about instant messaging - less about the protocols and the issue of AOL vs. everyone else and much more about how we efficiently manage a technology that has a lot of potential for being extremely useful and a lot of potential for making noise in people's lives.

Can you describe the IAB's relationship with ICANN?

Sometimes we look at ICANN in complete horror, and then we realize that all of the alternatives probably range between worse and much worse. We're quite happy to leave most of the policy questions to ICANN that ICANN is taking on.

Figuring out whether a new top-level domain should be allocated and what it should be called is one of these areas in which the IAB and the IETF have opinions but no competence. We will give advice to ICANN on some of the ideas that we consider dangerous to the way the Internet operates. We have a pretty good working relationship about 90% of the time and a serious headache the other 10. To paraphrase a colleague of mine: It's a source of continuing amazement how many politicians, committees, meetings and organizational structures it takes to replace one very competent and thoughtful engineer.

One of the hot buttons in the IETF is the issue of migrating from the current IPv4 to IPv6. Where do you stand on IPv6?

I understand IPv6 at a 10,000- or 30,000-foot level. I am statistician enough and data analyst enough to be able to read the handwriting on the wall about Internet address space exhaustion and to have my own predictions about how quickly that's going to happen. But IPv6 is not a protocol that I could design. It is not a protocol that I made significant contributions to. I can tell you it's really important. And I can tell you that all the other solutions are either worse or nonstarters. I am certainly an advocate of IPv6.

You've been involved in Internet research and protocol development for 30 years. Not many people can say that. What is your view of the Internet's promise today compared to what you thought 30 years ago?

There's a myth that nobody anticipated any of this growth or any of this penetration 30 years ago. Not only were we anticipating it, we were discussing it. So from that standpoint, we're on track. I really want to see this as a communications medium and an enabling one. And the e-commerce thrust fits right into that picture.

But I'm concerned that the Internet could become a one-way distribution medium for entertainment for pure economic reasons. And if it does that in a way that squeezes the communications aspects out, that will be very disappointing. If it happens in a way that spreads the network and increases the resources available to it, that will be wonderful. Which way those pressures will play themselves out is very hard to predict.

How might the IAB change under your leadership?

It's hard to say. I am the first IAB chair whose principal technical interest and background is applications rather than deeper-level transport or routing. That means I'm a little closer - just by background and instinct - to some of the hotter topics of today such as telephony, security or policy issues.

There are topics that come up at the fundamental technology level of the Internet that I can participate in as an educated lay person, but neither I nor anybody else has any delusions that I'm a serious expert. In that respect, a number of things are turning into a more collegial effort within the IAB because we're increasingly having to recognize that no one person has all the necessary expertise.

One of the criticisms of the IETF is that it has become too slow at developing Internet standards. Is that a fair compliant? What can you do in your role as IAB chair to improve the situation?

That's a really hard question. As a consequence of these scaling issues and growth, everything is getting more complicated. Some of the people who are complaining about us being slow are expressing a wistfulness for previous generations where they perceived things as easier. When we were a smaller community, if we managed to reach consensus on something, people would go off and implement it and that was the end of the problem. We're dealing with far more factions and influences now. The complications, the scaling problems, the internationalization problems and the security issues-we just weren't worrying about them nearly as much 15 years ago.

The second group of complaints about us being slow usually takes on the character of: "I've got this brilliant idea and it's fully worked out and the only thing I need in order to make it a commercial success is for the IETF to endorse it and do that real fast." We're certainly hearing versions of that a lot. Ultimately, most of us are convinced that admitting [protocols] of good quality that will be widely accepted in the Internet is more important than admitting something so quickly that we can't keep the quality up.

The other answer to this question is that we're probably taking on too much work. That gives us the choice between doing some things a little faster and more efficiently at the price of doing some other things not at all. One of the things that I will probably do as IAB chair is suggest there are certain kinds of work for which the IETF doesn't bring any expertise and as a consequence we should simply not touch it at all.

What kinds of work shouldn't the IETF touch?

There have been endless and frequent attempts to suck the IETF into legal and political policy areas. Every time the IETF touches one of these things, we succeed in making a demonstration of how poor we are at it. With the Internet wiretapping issue, the IETF made its conclusion that there are a lot of very complicated policy issues here which we are not very good at. The right thing for us to do is to continue to develop technology on an internationally unbiased basis and then let others work out the policy issues as appropriate.

How many hours a week does your IETF-related work consume?

It depends on the week. On a bad week, I feel like it's about 95….There was a particularly bad two-week period last October when--just with e-mail alone--I managed to hit over 2,000 incoming messages a day on average. This is the ultimate scaling problem with the Internet: there are more people who think you are entitled to their opinions.