Driving Innovation in Large Companies - A conversation with Raj Yavatkar, CTO Juniper Networks

This article contains lightly-edited excerpts of a conversation between Rahul Abhyankar, host of Product Leader’s Journey, and Raj Yavatkar, Chief Technology Officer of Juniper Networks. It covers Raj's journey of driving innovation at Intel, VMWare, Google Cloud and Juniper Networks.

Let's start at Intel and VMware. You were a Fellow, which I understand is the highest technical role inside a company. What does that mean, and how do you intersect with other organizations in the company?

When I went to VMware, there were very few Fellows. There I think they wanted to recognize people who were already industry leaders, having impact and how they could help VMware. So it was sort of different criteria, different role.

In Intel I was an individual contributor, so you have to influence across the company to get what you want. You don't have the resources directly available to you, whereas at VMware I was given an opportunity to do a startup within VMware. So that's a very different experience, where I owned architecture, product management, engineering. But Intel was really, it was cross-group influencing and continuously convincing people to invest in something that you believe in. So that was a very different role.

Back in 2010, I remember seeing a white paper about Intel's vision for cloud computing by the year 2020. They had a 10-year vision for cloud computing and Intel's role in it. How does a 10-year vision get built up within the organization? How did that come about?

The idea was to always paint a 10-year vision, understanding fully well that you may not be able to realize it, but to at least analyze what are the challenges you're likely to face and what are the opportunities. I always found that exercise very useful, especially if it's done cross-functional. It's not just by technical people. There are some product people.

Intel was investing heavily, starting in 2004, in the area of user experience. These were social scientists who went and spent time in different places, for example, to understand the PC's future. They went and spent time in remote villages of India or other Asian countries, to Africa, to understand what are their requirements. They also were part of such a vision creation process, which was a very powerful idea in my opinion.

Initially many of us, including me, were skeptical. You're talking social scientists. They have no idea about technology, so you know you're talking to people on a very different plane. But over time I came to appreciate what they bring to the table. They thought of things that we never thought of.

Fast forward from there you had the opportunity to do a startup within VMware. Tell us about that experience.

Raj: So in 2013, VMWare had already tried to build its own public cloud. It was not successful and people were very worried about VMware because VMware did not have a cloud story. So when I joined, coming from Intel, people were a little bit skeptical. Is this a software guy? Is this a chip guy? If you come from Intel, everybody thinks you are a chip guy. And I'm a distributed systems networking software guy. So my bosses, Raghu, who's now the CEO, and Ray, they were two in a box.

Ray especially told me: "Raj, you are a new guy, you have a clean slate. Why don't you go and figure out what should we do with cloud?"

I took the challenge. One of the advantages VMware has is the customers really like it. So I went on this tour to visit customers. I went to Minneapolis, St. Paul, Twin Cities. There I met so many VMware customers, about 10 of them.

I spent a week there to understand the problems. VMware had defined a new vision called software defined data center. The whole data center is software defined and we have seven products in network virtualization, compute virtualization, storage virtualization, all monitoring, analytics, all that. All those seven products had to be integrated to create a private cloud and customers complained that it was too much. They had to pay millions of dollars for professional services, three to six months to bring it together.

So I came back and wrote a five page white paper that we should all automate with the vision to bring up the private cloud in less than a day, down from three/six months, that's a pretty audacious vision.

And Raghu said: "Okay, you can go try. You can take four or five people and create a POC." This happened in March, April. By 1st week of December, we brought the same customers down from Minneapolis to Palo Alto and showed them two racks full of servers. Everything could be brought up automatically and that demonstration was just to gauge the reaction that we can build private cloud. It was all put together with toothpicks and gum.

It was like a prototype demo, but they liked it and they told Raghu, who was the Chief Product Officer that time, that yes, this is something that will really help. That was when he gave me the funding. So at that point it became a startup, because I had architecture, product management, engineering. That turned out to be my best professional experience.

And now that product has more than a billion dollars in revenue. It's the only organically created product outside of vSphere, where the company was founded. It was a very good experience, very different.

In order for product managers to be successful in collaborating with engineering, what are the attributes product managers need to have? How technical do they need to be in being able to have credible conversations with super technical people, engineers, Fellows like yourself?

The best product managers I have seen understand technology and they interact with engineers at their thinking level, at the same time making sure they abstract out things they want.

That's the most important thing.

I see nowadays product management is a discipline where people enter from engineering, marketing, sales, and you see different characteristics of people. Sales engineers becoming product managers and engineers becoming product managers tend to have a technical background. I've seen some people coming from marketing who take time to sit down and play with the product, play with the technology to understand that. They have a lot more credibility with engineers.

The word collaboration can be interpreted in multiple ways, because it's not just question of making people feel good and interacting with them, but instead collaboration means that you all truly create a working group in which people bring the ideas and you brainstorm.

And you have to bring people from sales architecture, sales engineering, regular engineering and product management. You have to. I find the input coming from the sales engineering side very useful. People think they are very short term focused. That's not true. They see the bigger picture many times as they're interacting with the customers.

The best example I have is Jay Chaudhry. He is the Zscaler CEO, a friend of mine. He came with that background and has built some really successful companies.

In your role today as the CTO of Juniper, how do you engage engineering and product? Product does not report into you as the CTO, so that's a separate organization. So how do you then bring that together with you and the Chief Product Officer, engaging your organizations into this long-range planning as well as incubation for new ideas?

You have to start with thinking about go to market.

Is it going to have a bigger revenue opportunity? For me, another criteria is software. We are trying to increase the company's ARR revenue, software revenue.

Once we do that, I assign two to four people to an idea to try it out, to give them three, four months to come back with something. If that fits the requirements, then we make it a Pathfinding project with actual number of people assigned to create what I call Minimum Viable Demo.

It's an idea based on MVP, but it's not quite MVP. But it's not a science fair demo that you throw across the wall. It's basically something that you can actually deploy in a customer environment to do a pilot. So it has to be robust enough and that is used to collect feedback before we decide whether it can go into Business Incubation or not. Business Incubation will try to make it a regular product with sales, revenue and stuff like that. That's a decision I make with my CPO. That decision is not mine.

First phase is all mine, my resources. I can assign them. But Incubation phase requires his buy-in because he sees now, aha! this can be built into a new revenue generating product.

The term Minimum Viable Product, the MVP, has been so overused, abused, burdened. How did you come up with Minimum Viable Demo?

The best thing is to go to customers. We have to ask them to try it out, but they are not going to try it out unless it's robust enough. So that's how this concept of MVD came about and people have liked it. So far I've gotten really good responses from that. Recently we did a pilot for a completely new product I'm building at Equinix. They provide a colo serviceand their feedback created some credibility to be able to take it to the next level.

This has been a fantastic conversation, Raj, really insightful. Thank you once again for taking the time to join me in this episode.