Jonathan Rosenberg chats with Mio – part two
Mio recently talked to Jonathan Rosenberg, the father of the Session Initial Protocol (SIP). SIP is the communications protocol that enables everything from video conferencing to instant messaging. In October, Jonathan stepped down from his position as VP and CTO of Collaboration at Cisco where he set the direction for Cisco’s collaboration strategy. Jonathan’s work in the telecommunications field over the past three decades has revolutionized the way we all communicate and he is well known for his ability to forecast technological market transitions years in advance.
In this second part of a two-part series, Jonathan Rosenberg shares the stories behind the creation of SIP, how a single conversation changed his life, and several other pivotal moments that transformed the telecom landscape. You can catch the first part our our interview here.
SIP 20 years later
Mio: It’s been twenty years since you first published the first core document for SIP, RFC 2543. When you first shared it online, did you have any idea how significant its publication would be?
Rosenberg: Absolutely not. SIP has been successful way beyond my expectations, and I have to believe that’s true for so many other things. Do people work on something and know it’s going to change everything? Now, pretty much every mobile phone in the world runs on SIP. But, in the beginning getting any market adoption was a fight.
“We had no idea SIP would be world-dominating.”
Mio: What were you up against when you were fighting for SIP market adoption?
Rosenberg: In 1999 VoIP was still seen as a toy. There were lots of naysayers who didn’t think SIP would replace the prior generation of technology. At the time, I worked at Lucent Technologies, a huge telecom equipment company, and Lucent made all of its money on traditional circuit switch technology. Here I was, this young renegade researcher saying, “No. I think we need something different.” There was a lot of resistance to that.
One of the pivotal moments for me was when my boss at Bell Labs Research at the time, Vijay Kumar, called my office. He said “Listen. I got a call from the standards people at Lucent. They were upset that you were promoting this new technology when, you know, the official position is we’re supporting this other technology.” And I’ll never forget he said, “I just want to tell you, great job. If you’re not pissing people off as a Bell Labs Researcher, you’re not pushing the envelope enough.”
That conversation energized me. As a young guy at Bell Labs Research, I didn’t really care much about what the official position was. I knew this technology was better and so I spent a lot of time talking to people at Bell Labs and working with product teams to convince them to add SIP support to their products.
Mio: Would you say collaboration is one of the threads throughout your work?
“You can’t do anything all by your lonesome self. All the things I’ve done were working with people who light the direction that I wanted to go and I would work collaboratively with them.”
Ultimately, a lot of other people did the work in developing software and building hardware to make this technology successful.
Mio: Was there a watershed moment when suddenly SIP was accepted by the industry, or was it just step-by-step process?
Rosenberg: It took years of additional development and maturation for SIP to get to the point it is today. But, along the way, there were also pivotal wins. This guy named Henry Sinnreich was this bleeding edge technology guy who couldn’t wait to get his hands on the latest and greatest technology. He latched on to SIP and became one of our chief evangelists and promoters. He used his influence at MCI WorldCom and got the attention of none other than Vint Cerf, who is known as the inventor of the internet.
There was a meeting with Vint Cerf, Henry, and a bunch of Lucent sales people. And then there’s me, this twenty-something year old researcher at Bell Labs Research. The whole conversation was about why SIP was the future and why MCI needed Lucent to support this technology. Vint Cerf was already pretty famous at the time. Getting his early support to push SIP made a big difference. All of these vendors like Lucent who wanted to sell to MCI Worldcom added SIP support to their products.
Light bulb moments
Mio: You’ve mentioned big light bulb moments for you during your career. What are the moments that were most pivotal for you?
Rosenberg: When we published RFC 2543, it had some bugs and we knew we needed to do a revision pretty quickly to mature the technology. But I was sort of dragging my heels on it because it was a big piece of work. This guy Christian Huitema who was another early pioneer in the field came to me and said, “This SIP thing is going to fail unless you fix it.”
Christian’s comment made me thoroughly believe that SIP required this level of massive effort to get it going. Otherwise it would fail because the technology wasn’t mature enough. The conversation with him triggered me to recruit a new editorial team to build a new version of the SIP spec. The work took two and half years. When we published, it was 269 pages and the longest document ever published by the Internet Engineering Task Force.
Mio: Was Christian right? If you hadn’t gone back and wrote the new spec, would history be different?
Rosenberg: Without a shadow of a doubt. RFC 3061 was an extremely good piece of work and it hit the mark on everything. SIP is a collection of documents and 3261 was the core. We published over a thousand pages of material at the same time. That made a big difference. It was in the beginning of the 2000’s when this technology flipped from “We’re not sure if it will be successful” to “Okay yeah this made it.”
Mio: Can you tell us about the work you did at dynamicsoft? What led to its acquisition by Cisco in 2004?
Rosenberg: In 1999 I had another one of those pivotal moments. During the dot.com boom, I was just a researcher working at Bell Labs struggling to convince anyone to build this SIP stuff. In the meantime, multiple startups were actively making SIP products and I knew them all. They all started to offer me jobs.
I went for a walk with my boss at Bell Labs, Vijay Kumar. The same guy who was happy I was pissing people off. During this walk he says to me, “Jonathan, there’s so much opportunity happening right now. There’s so much money in this internet technology community. Why are you still here?” So I went to work at dynamicsoft.
We had success with IMS (IP Multimedia Subsystem) and sold it to cable operators. At this point, wireless companies decided on SIP and the cable operators were also defining their next generation standards. They would standardize the systems so that companies like Cisco could build products they could sell to these cable and mobile operators. They would have multiple vendors they could buy from and the products had to be able to interchange with each other. So the standards are really important to this ecosystem.
We sold IMS to the cable operators to replace their prior generation of their technology. We worked successfully with Comcast, one of Cisco’s big customers. Since Cisco was worried they were missing out on this IMS thing when they didn’t have the technology, they bought dynamicsoft.
Mio: When were the standards for the mobile networks chosen?
Rosenberg: Around 2002. The standards body called 3GPP decided that SIP was going to be the future of the mobile phone network. This was another pivotal moment. We fought really hard for that. The mobile network operators needed all this technology to power the mobile network. I worked like crazy with a lot of people at standards bodies to define these technologies that would be used by mobile network operators. It was a huge victory and furthered the mind sharing momentum that this technology had.
Mio: How are standards for the entire telecom world decided? Is it a dozen people sitting around a table? Are there thousands taking part in this kind of decision making?
Rosenberg: This is the weirdness of how the IETF works. It’s a large group but it’s not a formal organization. It doesn’t actually have a membership. It has meetings. About one or two thousand people attend those meetings, and then it breaks into groups. Some groups will have hundreds of people while others might be less than 10. The decision-making is difficult.
Mio: Does the IETF work as a democratic system?
Rosenberg: No – there’s no voting. In a democracy, you register to vote, you vote, they count. Majority wins. That isn’t the case with the IETF. Someone gets appointed the chair, then when they want to make the decision, they take a hum, believe it or not, or they’ll say, “All those in favor, hum,” and people go, “Hum,” like literally hum. And then they’ll say, “All those opposed, hum,” and then other people will hum, and by volume of the sound, they decide.
Mio: Twenty years have passed. A lot of what you developed is still being used.
Rosenberg: It’s been great but it took a really long time for things to get adopted. We sold one of the first IMS products to Sprint in 2002 and they literally just turned it on for all mobile phones like two or three years ago. That’s another lesson of my life; big change is harder and takes longer than you think, no matter how cool the technology is.