How A Rocket Took Off: Juniper Founder Pradeep Sindhu (Part 5)

SM: Did segregating the problem set into a control plan and data plane slow you down as you prepared to go to market?

PS: No. It actually proved beneficial to split the problem into the control plane and the data plane. The software does not need to worry about real time. Software moved packets the way routers used to be built. That code path needed to be very tightly optimized. The more features added meant even more optimization. Writing real-time software where the structure of the software remains clean over time is almost impossible.

In our case, the forwarding was taken care of by hardware. Software did not need to worry about real time thus it could be written with a very clean, modular structure. This has made an enormous difference over time. Features can be quickly added without impacting real time. That made a very clean separation between the hardware and the software.

SM: Who was the first carrier to adopt your technology?

PS: Our very first $6 million order was to Cable & Wireless. UUNET was a very, very close second. Mike O’Dell at UUNET was absolutely instrumental providing the support and validation that what we were trying to do was right.

SM: How did you get to Mike?

PS: Dennis knew him personally. All of the early guys we hired were key players. They knew all the technical folks at these router and carrier companies.

SM: You delivered it in 1998?

PS: The machine powered up in April or May of 1998. We shipped the first machine in August.

SM: At that point you then gained reference customers so you could go on to other key people.

PS: Exactly. UUNET was the largest Internet service provider, and they were very high profile. The problem at that time was so acute that there was a giant sucking sound. We had no PR yet people were calling us non-stop. We could not make units fast enough. In the beginning we did not need a marketing side of the company. Our machine made the network inherently stable.

SM: How much did you sell that machine for?

PS: The benchmark had been set by Cisco but our machine opened a new category. A fully loaded M40 machine was around $1.5 million in full OC48 configuration.

SM: What did you revenue ramp look like?

PS: In 1998, a partial year, it was $6 million. In 1999 we had $100 million and 2000 was $400 million. In 2001 we did $900 million. After the M40 we rapidly built the M20 with the same technology. Since the M40 had stabilized the network the demand for bandwidth went up much faster. OC-48 was 2.5 gig and there was a need for OC-192 which was 10 gig in 1999. Our project was to do a 8x OC-192 using the same base technology that we already had. We were the first 8 x 10 gig router. People before had said that routers could not outperform switches, and yet we brought a full 10 gig router to the market a full two years before a 10-gig switch.

We very rapidly came out with other variants as well. In 2000, we started work on the T series. Service providers had told us that they needed the same capabilities at the edge of the network. We worked on variations of the machine and enhanced the technology to support features found at the edge. From there things caught fire.

Our customers were still carriers. That is a small group that does very large deal sizes, they knew us well, and we did not have to work hard to sway an entire market. Of course one false move and it’s over.

SM: When did you go public?

PS: We went public in June 1999. I was CEO only for the first 9 months of the company. In August 1998, we brought in Scott Kriens. When I had first approached the VCs we had discussions about being CEO. I told the VCs that I thought my concept was going to be very important and that I did not have the background to be the CEO of a very large technology company. I loved technology so I needed their help to find a CEO. We started our CEO search in March and interviewed six or seven candidates. In August I met Scott, and he joined in September.

SM: What made him the right fit?

PS: A couple of things. Scott’s skillset is complementary to mine. He understood the industry. He knew frame relay and had been the VP of sales for a company Cisco had just bought. He also had very high integrity. I knew that was a decision that, if wrong, would have been a disaster. I had the help of the VCs. There were 17 of us in the company and every single one of them interviewed Scott. I talked to people who Scott had worked for, people who had worked with him, and people who reported to him. I tried my absolute best to get a 360-degree view. We unanimously agreed that Scott was the right guy. He was CEO through 2008.

This segment is part 5 in the series : How A Rocket Took Off: Juniper Founder Pradeep Sindhu
1 2 3 4 5 6 7