Next week, Randy Frank will take over as the new CTO of Internet2, an Ann Arbor, Mich.-based consortium of 200 U.S. universities conducting advanced networking research. Frank joins Internet2 from Fidelity Investments, where he served as chief technologist for the financial services firm’s Center for Applied Technology. Before that corporate post, Frank ran high-performance computing centers and research networks for the University of Michigan and the University of Utah.
Now, Frank is looking forward to getting back into academia, where he can get involved with cutting-edge Internet research in such areas as dynamic provisioning of high-bandwidth circuits, distributed authentication schemes and IPv6, an upgrade to the Internet’s main communications protocol. Here are excerpts from Frank’s conversation with Network World national correspondent Carolyn Duffy Marsan about his plans at Internet2:
Why are you taking the job as CTO of Internet2?
My background prior to Fidelity was all in academia. I started my career there, and I enjoyed it. After 10 years at Fidelity Investments, I decided for a number of reasons that I like the academic environment better. One reason is that I like talking about what I’m doing. At Fidelity, it was an unusual case where we got permission to talk about what we were doing. My background is in being able to build experimental networks. I really wanted to get back to that kind of environment. At Fidelity, the pressure was on to keep the trains running on time and keep the production networks up.
What do you see as the significance of Internet2 here in the United States?
If one goes back to the history of the Internet and Arpanet, in the late `80s, it was an academic environment. The University of Utah was one of the four original Arpanet sites. In the `90s, as the commodity networks took over, it was time for the academic and research community to move on to something new. Internet2 was really driven by the fact that we still needed a network we could do experimentation with. I’ll give you a good example: multicast. IPv6 is another example. We needed a network where we could experiment with all the technologies that we’re no longer able to drive adoption on the commodity network. The campus and regional networks combined with Internet2 is the national fabric for pushing ahead with new directions in networking.
What is the position of Internet2 globally in comparison to large, well-funded research and education networks in Europe and China?
Internet2 peers with leading organizations around the world in terms of building a worldwide research network. One example is our Dynamic Circuit Network (DCN), which is now establishing dedicated high-performance links by partnering with sister networks in Europe and Asia.
What new initiatives do you plan to launch at Internet2?
It’s a little premature to say, but there are a number of initiatives that are going to continue on at Internet2. I’m a strong proponent of multicasting. If one looks at the pressure on the commercial Internet being able to sustain video transmission, I think a lot of our experience with multicast is useful. We have a multicasting-enabled Internet2. We can send a single stream across the Internet and not replicate anything unnecessarily. Today on the public Internet, with the inauguration of Barack Obama, the network didn’t collapse but a lot of people had a less-than-great experience. Multicast is a technology that solves that problem brilliantly but is unavailable on the commercial Internet. It’s been the experience of Internet2 in taking these technologies like multicast and showing their validity.
What role do you see Internet2 playing in the deployment of IPv6?
From my experience at Fidelity, it’s clear that IPv6 at the moment is not seen as something that adds value. In academia, network address translation (NAT) was always viewed as a bug. In industry, NAT is viewed as a feature. The fact that all desktops aren’t publicly addressable adds to security. Internet2 needs to demonstrate how you can have full end-to-end connectivity and still have security. Most of industry is asking why we should upgrade to IPv6 when the current network is working fine. Organizations like Intenet2 need to increasingly talk up the advantages of IPv6. One advantage to IPv6 that you don’t get with NATs is that when you do mergers and acquisitions, you won’t have multiple people using the same address space.
What other emerging technologies are you excited about experimenting with at Internet2?
DCN, which is in a very early stage in Internet2. If you need high bandwidth capacity today, you’re forced to put in a dedicated circuit. What’s happening with DCN is one can build a network layer that allows you to get a fair amount of dedicated bandwidth on a very dynamic basis. A lot of companies like Fidelity are looking at very high-resolution conferencing, but to date all of these telepresence systems run across dedicated networks. You have to put in a T3 or greater because running that over the public network is really impractical today. With DCN, you can build a network where one is able to dial up -- figuratively speaking of course -- a sustained amount of bandwidth for a short period of time without the expense of dedicated circuits. Internet2 is very much on the forefront of that.
One of the huge successes of the Internet2 community is addressing the whole question of distributed authentication. Our middleware technology Shibboleth is the first large-scale distributed authentication environment that is succeeding. It’s an architecture that allowed each campus to do authentication the way it always did. It builds a model to allow cross authentication, so campuses can trust the authentication model of other campuses.
What do you see as the key challenges facing Internet2 in the future?
One of the management books I love is “The Innovator’s Dilemma.” It talks about figuring out when it’s time to do the next thing. One challenge for Internet2 is figuring out when it’s time to move things into the public Internet and then move on to the next thing.
Given your experience at Fidelity, how do you plan to improve relations between Internet2 and the U.S. business community?
Where I will get the most out of spending 10 years at Fidelity is in understanding where Intenet2 can meet the needs of not only academia, but industry as well. Going to Fidelity gave me a very different appreciation for production networks. It wasn’t until I went to Fidelity that I realized I didn’t have a clue about scale. People talk about five nines of availability, but at Fidelity the goal was 100%. The goal was to architect an environment so that no single points of failure existed, and with redundancy so that no failure should ever cause a customer a visible outage. Getting to that level of redundancy at a place like Fidelity was absolutely justified given the business it is in. But you’d have to question whether that was the best use of money at Internet2. Fidelity gets reliability by being stunningly redundant.