Howard Frank Looks Back on His Role as an ARPAnet Designer

April 25, 2016

In 1969, Internet Hall of Fame inductee Dr. Howard Frank co-wrote a proposal to design the network structure for the ARPAnet. He didn’t know it at the time, but this work would cement his role in Internet history. Dr. Frank spoke recently with the Internet Hall of Fame about that project, his work in applying the technology more broadly, consulting for the White House, and what he thinks of the Internet today.

IHoF: What was your role in the development of the Internet?

HF: I formed my first company, Network Analysis Corporation (NAC), in 1969. One of our first contracts was with DARPA (Defense Advanced Research Projects Agency). The topological design of the ARPAnet was, at that point, for four nodes, and they were all on the West Coast. Additional nodes had been ordered, but there was no scientific design. We were hired to do that because we were experts in design.

IHoF: How did you become involved?

Up until 1968, I had been a professor of electrical computer sciences at Berkeley; my specialty was networks. I gave a speech in Washington D.C. on the vulnerability of communication networks and someone in the audience heard me. Then I got a call asking me to become a consultant for the Office of Emergency Preparedness in the Executive Office of the President.

They asked me to come spend a year on leave. I accepted, and in the summer of 1968, my wife and I drove to Washington from California.

I set up a little group of people who were government employees and consultants. We were told, “Do something useful,” something that would show network analysis could be useful in a White House context. We worked on two things: network vulnerability and network design.

We found a real problem at the Federal Power Commission (FPC) with the design of offshore natural gas pipeline networks. So, we came up with a way to improve applications. It gave the FPC a tool to evaluate proposals for gas pipeline network construction in the Gulf of Mexico, and saved several hundred million dollars. We demonstrated it in 1969 and it turned out to be fantastic success.

While I was in Washington, a friend introduced me to Larry Roberts, who worked with the program manager for the ARPAnet. We had dinner one night and he talked about packet switching.

After I formed NAC, I thought of Larry and the conversation we had and I went to visit him in Washington. He said, “I have this design, but I have no idea if it’s cost effective or whether it carries the right information. Could you folks look into it?”

We came up with a way of analyzing the network and a computer program for designing the network. We showed him an alternative design with 40-percent more throughput, at a 25-percent lower cost.

He asked, “Could you become the design consultant for us?” He wanted to consider factors of reliability, and started asking about how this technology would scale up. Could it be grown to serve very large networks?

Computing was in its infancy at the time; everything was a real effort. It required a lot of different technological smarts. So we added specialists and built a much more comprehensive computer design.

IHoF: How did the project progress?

HF: By 1974, we were designing for reliability, throughput costs and developing profiles for how this technology would compare to the telephone technology of the 1970s. The packet-network technology was cheaper to carry the same amount of material. It was much less expensive.

By that time, Bob Kahn had come into it. He posed the question, “How would the technology (packet switching) look when you compared it to the telephone circuit-switching technology of the day? How would it look for the entire Department of Defense (DoD) communications network?” This turned out to be a two- to three-year project.

We worked with the Defense Communications Agency (now the Defense Information Systems Agency). They supplied us with databases, requirements and locations of the various networks of the DoD. I was the principle investigator for the project.

This was a huge network. We had to invent whole new technologies and literally design packet switches for carrying voice and data. Then we wrote our report, which became a well-known series of studies.

It turned out that packet switching was better for voice, data or both. The results were reviewed by Bell Labs, and these were the first studies of the network technologies we have today.

IHoF: What part of the work did you find most challenging?

HF: There were a lot of pieces that were challenging. No one had built a packet switch for voice. Building a model for a massive network that had to be designed from scratch required a breakthrough in network-design technology.

Larry and I made a presentation to AT&T/Bell Labs about how this packet-switching technology applied to their problems. We showed the cost effectiveness of the technology compared to the circuit-switching technologies AT&T was using. The packet-switching technology was much more cost effective. Larry talked about the network and I talked about the design. It turned out they weren’t interested in using the technology.

Some of the work we did had to do with packet radio technologies. Larry and Bob asked us to join the Packet Radio group. Packet Radio was a project that DARPA was investigating to rapidly establish communications in hostile environments.

Our time as contractors with DARPA was from 1969 to 1977.

IHoF: When you gave your Internet Hall of Fame acceptance speech, you said it was “a great pleasure to see what has taken place over the last 20 years.” Can you tell us about some of the communication advancements that you consider among the most important?

HF: In the 1980s, the world was dominated by circuit-switching technology. There was no “Internet.” It didn’t exist as a separate technology until the 90s. You had to dial in to a local telephone exchange. There were no real data communication capabilities anywhere on Earth, except in the military.

Back then, whatever you did was hard and unreliable with very little capability. But, it kind of worked. Then, you got it to work better, then faster and then reliably. It was hard to pull it all together. 

When we saw the Internet was carrying more commercial traffic than government or university traffic – billions of people – that was an amazing transformation.

The next phase in transformation was when basic telephone technology disappeared and was replaced by packet technology, which now allows voice, data and images.

Now, everything works together quite well with immense speed. We move megabits around the way we moved bits around before. Netflix? That wasn’t even possible 10 years ago. There has been enormous transformation for the communications infrastructure. It’s changed the course of human history.

IHoF: What do you think of the performance and reliability of today’s Internet? 

HF: Today’s Internet is highly reliable compared to the Internet of 25 years ago.

IHoF: What's your view of social media platforms such as Facebook and Twitter?

HF: To be honest, I don’t use any of them. I don’t want to open my social life to an unconstrained, outside world. I’m a bit of a conservative in that regard. 

I don’t use ATM machines either. I don’t trust the security of the networks. You can’t tell what will be compromised next. The bank computers are not 100-percent secure. The links to the ATMs are not 100-percent secure. The Internet itself is not 100-percent secure. It was not a particularly important issue when it was designed in the 1970s. It was an add-on.

The weakest link in a network is the end point – people on their computers.

IHoF: What projects or adventures are next for you?

HF: I retired as of last November. I’ve been bodybuilding ever since. It’s my new project.




Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
By submitting this form, you accept the Mollom privacy policy.