Feature Story

Interview With Jim Mitchell

25.Jun.02--Sun.com recently sat down with Jim Mitchell, Sun Fellow, vice president, and director of Sun Microsystems Laboratories in Palo Alto, California, to learn more about how the Sun Labs came to play such a pivotal role in Sun's culture.

Q: Hi Jim. First, if you could say one thing about innovation at Sun Labs, what would it be?

A: There are many things, but one thing I think it's important to discuss is our commitment to process innovation, not just product or technology innovation. That is very important. When you innovate a process, you get immediate value out of it--as opposed to waiting through a product development cycle. Better processes save significant resources--time and money.

Q: Can you give me an example of critical tools and/or processes that you're talking about?

A: Sure. Phil Yelland's [senior staff engineer, Sun Labs] work in demand forecasting. Demand forecasting is the toughest thing a company does and very tricky for a manufacturing company. It offers a huge advantage in ordering parts--you don't want to overshoot or undershoot. If you don't forecast correctly, and you have to buy a component at the last minute to make your inventories, you'll pay a premium for any of those parts. And of course high premiums mean your margins go down. The better we do demand forecasting, the more money we save. If you overestimate demand, that costs money. If you underestimate, you're not able to take advantage of the demand that's out there. But if you avoid underestimating demand you make money.

Q: Do you have metrics or numbers on what Sun has saved using this demand forecasting technology/methodology, or is it still new?

A: It's being implemented as we speak.

Another Labs' project is developing a process called Project Ace that can help Sun immensely. Project Ace is a tool that helps folks build enterprise Internet applications easily. You get an improvement on the order of a factor of five if you write on Ace as opposed to on the [Java 2 Platform, Enterprise Edition] J2EE platform. This not only helps customers, it can help us with our own IT processes.

In a volatile world, ACE can facilitate immediate deployment of IT assets. The financial market is a particularly good example. New financial instruments are coming out all the time. And whenever a new instrument is introduced, the faster a financial house can implement and generate trade based on it, the more they'll be able to set themselves apart from the competition.

Q: How did you get started at the Sun Labs?

A: I joined Sun in 1988 to help Bill Joy [Sun chief scientist and corporate executive officer] do research on a project that was done in conjunction with AT&T in merging the System V and Berkeley (BSD) strains of UNIX®. That was project Spring, a straight research project--the Labs didn't exist yet. In 1990, Bill Joy and Eric Schmidt--now the CEO at Google--thought it was time for Sun to have a research organization.

At the same time, Eric thought we could merge with a consulting company, Sutherland, Sproull and Associates. That's where Bert [Sutherland, director emeritus, Sun Labs], Bob [Sproull, Sun Fellow and vice president, Sun Labs, Burlington, Massachusetts], and Ivan [Sutherland, Sun Fellow and vice president, Sun Labs] were. Bert had spent time at Xerox PARC. I think we called Bob "Mr. Brain" even back then. We put together a plan for what the Labs would look like--the mission, funding, etc., took it to Scott [McNealy, Sun chairman and CEO], and the management team, and they approved the creation of the Sun Labs.

The combined team became the nucleus of the Labs--me, Steve McKay [Sun vice president], Bill Joy, Eric, and the folks from Sutherland, Sproull and Associates, and Wayne Rosing, who's now chief of engineering at Google. So by 1991 Sun had a research lab.

Scott didn't want the Labs to grow too big, so we had to make sure we were good at "scooping" technology from the outside. If you're small, you can't invent everything--you have to pick it up and use it wherever you can. And a big piece of the charter was that we'd move the technology we create into the company for the benefit of the shareholders.

So you can boil it down to three things: One, you innovate--you make it up and you find it. Two, you demonstrate the technology; you have to show the technology to make solid decisions about it. And three, you have to deploy the technology.

Q: What's your background?

A: I'm originally from Canada and got my B.S. in physics from the University of Waterloo, then my Ph.D. from Carnegie Mellon. I'm old--I started college in 1962. It was different then. It's like they say: If you scratch an old computer scientist, you'll invariably find a physicist.

I spent from '70 to '84 at Xerox PARC. The Golden Age. Ethernet, bitmap displays, WYSIWYG interface. All that stuff came out of PARC at that time.

But I found the technology transfer issues frustrating. Apple and Microsoft took some PARC technology and they made a lot of money from what we developed there. Xerox didn't know what it had. So I went to Acorn--it' s the Apple of the U.K. in the education market. They were working on the Acorn RISC chip design, which became the ARM. That was basically a four-year MBA in the trenches. There were 230 people at the company. I was involved in every aspect--manufacturing, shipping parts--everything. It was a great real-world education. So when I came to Sun, that experience was helpful to understand both sides. You're straddling product and market realities on one side and research on the other. At Sun we believe that if you want to be good at research, you have to know about products. So we attack problems that help Sun's current business and we do things that will create new business for Sun.

Q: We hear about Sun Labs playing a role as talent magnet. Can you explain that?

A: We're instrumental in the flow of good people into Sun. One of the important roles we play is as recruiting arm. Sun Labs can recruit people into Sun that we'd have a tough time hiring straight into product divisions. We're a great source of outside-the-box talent. It's important that we participate in outside technical communities and academia to draw talent and to exchange ideas. So we publish papers, go to conferences, all the academic stuff, and we hire lots of undergrad and graduate level interns. This is all part of marketing ourselves as a research organization.

When we are ready to turn research into a product, some or all of the people will move into the product teams--usually three to five people. They have a great success rate transferring over, and we've helped to feed the talent pool on the product side. Some of the folks come back to the Labs after a product is completed, but the overwhelming majority stay with product teams, which creates the need to hire fresh blood into the Labs.

Q: Is that focus on turning technology into products that work a differentiator for Sun Labs?

A: Most good Labs do build things. Some, like IBM, also do basic research--physics. I mean, they run foundries, after all. If you invent a better transistor, and you own the foundries, you can make a better computer.

But IBM recently tried to increase the relevance of its research arm by having half of their funding come directly from product divisions; a tax, if you will. Sun, on the other hand, doesn't have to tax product divisions. The problem with relying on the product division for funding is when times get tough, you're high and dry. Sun leaves that decision to a high level so the vines that won't produce grapes for a few years get taken care of.

Q: It's recently been said that innovation isn't relevant to a commoditized market.

A: Hmm. It's not like growing corn or cattle. It's not that kind of commodity. Whoever thinks that is making the mistake of thinking there's no innovation left. For example, telephones were widespread 50 years ago. They were essentially a commodity even then. But cell phones have taken over in the last several years. Isn't that innovating in a commoditized industry?

Q: Where does Sun innovate where others don't?

A: One example where Sun can do what's difficult for others is in the integratable stack. We own that technology, so we are free to do things across the stack to make all the elements work better, tweaking the top part so that the bottom works better. Intel, on the other hand, is going to have a tough time telling Microsoft to change its OS if [Intel] finds a radically different way to build a processor.

You have things that are a commodity, but because of the innovation headroom that follows from Moore's law--and owning the stack--there's ample opportunity for innovation at Sun. People who say you can't innovate in commodity markets don't understand innovation.

Q: Can you tell me a little more about how Sun Labs fits into the overall ecosystem of Sun?

A: Absolutely. The thing to remember is you've got three segments of technology development. Imagine a left-to-right chart. On the left is basic research, which is universities and government research labs--that kind of thing. In the middle you have applied research. And on the right-hand side of the research continuum you have product development. There's innovation in this area, but it's done on a schedule.

Sun Labs' research focus is in the middle, the applied research area, where our project research is aimed at something that can be turned into something--a product. But at the same time, in learning new things, we learn other things which can be as important as what we were originally pursuing. That's the sweet spot for Sun.

It's hard to take basic research and turn it onto a product. That' s why our sweet spot is in the applied area. That's what I was referring to earlier when I said we had one foot in academia and one foot firmly planted in the real world of solving real problems. And being able to demonstrate a technology with prototypes is essential to bridging those two worlds.

Q: How have the Labs managed to stay focused and energized over two decades?

A: It's funny. If you look at the life cycle of a research organization, many industrial research labs come and go. They have a golden age that usually lasts about 10-15 years, then things go bad. The problem is they fill up at the outset with young folks who are full of enthusiasm and ideas. They are energized for a number of years, then they begin to grow old. Meanwhile, there is not a lot of room for a new crop of minds to replenish the system. Xerox PARC has gone through two such cycles. Sun Labs hires young blood and then exports it to the product development side, which allows us to renew ourselves. The average age in the labs at Google is, I think, 27. If you go to Xerox PARC, though, the average age is 45. That's not very healthy. Here, the average age is, I think, around 35.

Q: What about the role of the Sun Labs outside of Sun, in the world at large?

A: If you look at the Internet Engineering Task Force, we run more of their committees than any other company. That's not just a political game, it's a technology game, like the Worldwide Web Consortium. We generate 15 percent of Sun's patents. With our intern program, students get to see us up close and personal. And a number of us are on advisory boards of schools across the U.S. Bob and Ivan, for example, are on the National Academy of Engineering. Whit Diffie [distinguished engineer, Sun Labs] and Susan Landau [senior staff engineer, Sun Labs] testify before Congress. Whit is viewed by the military as "the" expert in encryption and security.

Q: Do you find it daunting that companies like IBM and Microsoft can just throw money at research?

A: My reaction is, hey, Microsoft has 200 people working on that problem versus the five we have working on it? Good. That's sure to slow them down. Remember, a few smart people working together as a team can change the world. Once you get up to 15-20, you've got management. And we all know how conducive that is to innovation.

Return to the feature.