Interview with Bob Hinden, Nokia Fellow


Date: May 7, 2006

line break image

Question: Bob, you attended the first IETF Meeting. How did you get involved in the IETF?

Bob: I have been involved in this work even earlier. I started working for a company called Bolt, Beranek & Newman (BBN) that designed, implemented and operated the ARPANET. I started working there in 1978. The ARPANET existed since 1969, so it was a fairly established packet switching network. I started as a programmer. BBN had built a multi-processor system called the Pluribus and I did a lot of software development for that which was also used on the ARPANET.

At that time TCP/IP was in its early stages and I worked on the project that added TCP/IP to the Terminal Interface Processor (TIP) part of the IMP. This involved replacing the original ARPANET host-to-host protocol, the Network Control Protocol (NCP) with TCP/IP. This was my first involvement in IP.

That means I wrote one of the first TCP implementations. There were about 4 different TCP/IP implementations at the time. That was a lot of fun and a real challenge. I was not involved in routers yet. We called it the more research oriented group. It is funny to say this now – in hindsight, the whole thing was research. We then got responsibility for building what we called gateways (we call them routers now). I realised that the routers were going to be packet switches in their own right. This was a radical idea at BBN which believed in the ARPANET style packet switching and was ultimately evolved on X.25 packet switching commercially. They thought IP and routers were not reliable. It was real hard to get them to take it serious as a business until it was much too late for them to be a big player in the router business. I was leading the group that developed and deployed 4 generations of gateways/routers, the HW, SW, operations etc. I don’t think we had any idea what it was going to grow into. Many of the problems we still talk about today, we already had then (e.g. that the routing tables are getting too big). Of course, the scale is different. We were thinking about what happened when we got to a hundred, not hundreds of thousands or millions. But if the router couldn’t do it, it was a big problem. That was in the days where the distinction between the hosts and the routers was less distinct. We then invented this idea of interior and exterior routing to try to separate them. There was an earlier protocol and then eventually BGP replaced that.

As part of doing this work at BBN – initially for ARPA and later for the Defense Data network – there were periodic meetings. It predates the IETF but evolved in creating the IETF as an activity to formalised the development of the protocols. Mike Corrigan was the first chair.

Later, when the IETF became bigger and Phil Gross (the IETF chair after Mike Corrigan) decided to break it into different areas, I became the first Area Director (AD) for routing, because I had done a lot in that area at BBN. During my time as the Routing AD, we created and standardised all of what are now the current routing protocols: RIPv2, OSPF, IS-IS, BGP. It is interesting to note that some of the WGs that started then are still going on.

Then I got involved in the IPng process and we worked with a team (together with Steve Deering) and developed what became the next version of IP: IPv6. That work has been mostly wrapped up now, we should probably close the WG sometime this year. I have been doing that for a long time.

But I have also been doing other things: I chaired the vrrp (Virtual Router Redundancy Protocol) (?) WG and worked with a team writing the protocol. Unlike with IPv6 which is trying to solve the big problems, it is nice to work to solve smaller problems sometimes (laughs). I think the protocol is very successful, it is deployed very widely and solves real problems.

When I first got involved in the network at BBN I knew it was going to be interesting, but I had no idea it was going to create what we have today. And I don’t think many people really knew then where we were going. I think that is actually good. Because, if we had been trying to design for today or even for beyond today, we probably would have made something very complicated and it probably never would have worked.I think it was the simplicity and easy interfaces of IP and TCP and that no tight coupling between the devices was needed, that has allowed the Internet to go where it is now. It allowed different companies, different organisations and different people building things. I believe this is the only way to create not just the technical side, but also the business environment that allows this to work: in a distributed way where there isn’t one person in charge.

Things that are very tightly coupled are very hard to scale. You can see that in the original way the telephone network started out. In the US and in many other countries, there was one organisation building the network, setting the standards, building the technology.

The Internet never followed that model. From the start it was very distributed and flexible. That is one of the reasons why the Internet has grown and is so diverse today.

Question: Let’s look into the future. You just said, the development of IPv6 is almost done?

Bob: We now need to do the operational transition which will be very hard. What originally caused us develop it, was that we were running out of IPv4 addresses. And yes, it looks like we will be running out of IPv4 addresses. It took longer than we thought, for a many reasons. It will be interesting to see if we are all able to get v6 turned on (it is actually fairly well deployed in a lot of systems). Everyone needs to think it is in their self interest to deploy it. We have the problem of the commons, I think.

Question: What do you want to work on next once this is done?

Bob: I am not sure yet. I have some interest in the manet WG and similar work around auto-configuration. I probably don’t want to work on anything that is quite as big as IPv6 again, at least for a while. I want to give it some time and think about what to do next.

Question: Do you think there are any big issues the IETF needs to address?

Bob: We still have the problem on how to make the routing work better. It was interesting to hear Dave Clark’s talk again on the big problems we have on the Internet now and 14 years ago. Already 14 years ago the big problems were routing and security! I think this is an indication that just because we know there is a problem, we don’t necessarily know how to fix it. We have been trying to make a more secure Internet for a long time and we have certainly done a lot of incremental things, but I don’t think anyone would call it secure. There is a lot of conflict between making it open and flexible and making it secure.

Routing has its own challenges. The problem we are trying to solve is a lot more complicated than it was. It is not just dealing with the number of routes. It is policies, people dealing with each other in companies that compete with each other. There are many different aspects that make this a much harder problem, and I don’t think we have any tools which can make this much better. So, there will be lots to do. I am not worried that we will run out steam. People are pretty good in making things better in an incremental way. Computers will continue to get faster and have bigger memories.

Q: What do you think will be the next big development on the Internet?

I think there are two different dimensions where we will see lots of growth: one is of course in places that don’t have the Internet today. Parts of the world that don’t have a lot of technologies or infrastructure. That is a big challenge in many ways. It is hard to have a router or a computer if you don’t have electricity.

The other dimension is the development to make everything on Internet interconnected. Instead of just having the laptops and PCs and servers we have to today, we will have an increasing amount of devices that are small or embedded. They may be self-organised but networked together.

In the first case the network is getting wider and covering more area and more people. But I think it is also going to get a lot more denser with more devices on the network all the time.

Building this big, complicated, very dense network where everything is networked together and everything has a potential to talk to each other and in a secure way, so it is usable, will keep us all busy for a long time.