Feature Story

Interview With Bob Sproull

25.Jun.02--Sun.com recently talked with Bob Sproull, Sun Fellow and vice president, who heads the East Coast Sun Microsystems Laboratories facility, located in Burlington, Massachusetts, to learn more about the Labs.

Q: Hi Bob. Let's start out by talking about how you came to Sun.

A: I had a background in computer graphics and IC design, with a Ph.D. in computer science. While I was doing academic research on the faculty at Carnegie Mellon, a bunch of us started a company (Sutherland, Sproull and Associates) and Sun bought us. Ivan [Sutherland, Sun Fellow and vice president, Sun Labs], his brother Bert [Sutherland, director emeritus, Sun Labs], and I were part of the founders of the Labs.

Q: What was the Labs' initial mandate?

A: Well, we had a kick-off meeting with Scott [McNealy, Sun chairman and CEO]. Scott said he wanted the Labs to be the eyes and ears of Sun. We knew that engineering groups are typically focused on solving immediate problems. So there's something called "absorptive capacity." This is the ability of a firm to recognize the value of new, external information, assimilate it, and apply it to commercial ends. It's crucial to any company's innovative capability. From our perspective, this is not only ideas, but people, so we wanted Sun Labs to be a recruiting magnet as well. One of the great things about the Labs is we don't operate in a vacuum. A lot of these people we recruit wind up transferring to product groups and heading up the development needed to bring a technology to market.

Right from the beginning we were intently focused on this idea. Now sometimes we'd bring people aboard and they'd look at an area where there was no existing need per se; but where we saw value down the road. You know, Sun Labs was never intended by Scott to be the source of innovation at Sun. The Labs are only one vehicle. We like to say we create technology options for Sun.

Basically, you have two types of investment--you have investment in the idea, where you develop it and build a prototype, and investment in the execution of bringing that to market. The way we approach it, we're focused on risk reduction. That's why we build prototypes of everything. Prototypes are very persuasive in a culture like Sun's. A paper presentation is not nearly as valuable in making your case.

That's another thing we've emphasized since day one: There are two ways that we can deliver value: Product innovation and process innovation. Bert Sutherland pointed out years ago that when you innovate better processes, your financial return is realized much faster. Inventory turns are a good example. You can save a great deal of money there right away.

One area where we're innovating is in the use of CAD. We use a lot of CAD and CAD tools, and we buy much of it from outside vendors. But sometimes, either the tools or the suppliers themselves aren't adequate to meet our needs, so we build our own. Sometimes someone in the Labs just has a "eureka" moment, where he gets a better idea on how to do something.

Q: A constant complaint with vendors is that technologies roll out before they're ready. Do the Labs ever look at products already on the market and devote resources to improving these things?

A: Well, it's not necessarily a bad thing to bring out products that aren't mature. You can have customers buy too soon, of course, and have that lead to problems. But Microsoft, for example, will deliver a product early on and then allow the market to decide how it can be improved. Until you get new products in the marketplace your data isn't going to be very good as to how to improve it. There's just no substitute for the real-world experience. So some of that problem is irreducible.

And you do modify and make choices you go back on. I remember in 1992, Sun decided ISDN was the future of telephonic data communications. So we put it into our workstations. But ISDN never took off. We realized pretty soon, that ISDN wasn't a good, forward-looking feature. So we took it out.

By contrast, Ethernet capability was just a plug-in module, no big deal. But Ethernet took off in the marketplace and quickly got hardwired into the motherboards. The reason that we changed our products in both of these cases--to drop ISDN and to harden Ethernet--was that the market dictated the technology.

Q: How do some of your innovations get started?

A: Ten years ago, networks were used extensively by our customers, but the Web hadn't happened yet. The network was pretty much a client/server environment, used for things like file sharing via the NFS (network file system) protocol. It quickly became clear that customers would store lots of data on file servers all over the network and would then have trouble finding it. So we hired Bill Woods to start a group to work on long-term technology to improve search and retrieval techniques.

Over the years, this group has contributed to several Sun products and processes. A technique it developed was used to improve the search capabilities of Sun's AnswerBook product, an early tool for browsing online documentation and manuals for Sun's software and hardware products. Later, a more complete suite of the technology was used internally by Sun's service organization to search its database of trouble reports and solutions, an application named SunSolve. Most recently, we converted the technology into Java, in the form of a "Java search engine" in close conjunction with an engineering group. They have used it in Sun's email server product to let customers search their mail folders.

So this is a story of an investment we made in a long-term technology vision, with no clear view of where it would lead or how it would be used. We bet that this would be an important problem, with ample opportunities for leverage. And it's had numerous payoffs along the way.

Q: What are some initiatives on the horizon?

A: Certainly one thing that we anticipate happening is that lots more stuff will attach to the Internet. Today it's your computer. But soon it will be buildings--for energy control, security, motion detection, things like that. Also, a lot of connectivity will happen in residences and cars. We'll see a tremendous proliferation of this kind of connectivity and new software, secure software, will be needed for this. How do we make sure that only the right people can get to information about appliances or cars?

And the continuing expansion of wireless devices. We're pulling together a project to look at how people get work done with mobile devices in conjunction with the wired infrastructure and services. It's relevant to Sun itself, because increasingly we don't have people permanently tied to offices. So more and more, the network will be the glue that holds it all together.

One economic incentive that hasn't flown in the past is general collaboration tools. They haven't done well. So we have to figure out how you do things that are specific to a particular collaboration itself, but this isn't necessarily going to be the same software for every customer.

Q: What are the Labs doing in the Java technology arena?

A: We have a fair amount of activity going on with Java technology. One thing we're working on is better garbage collection. The Java software does automatic storage management. If you need a hunk of memory to store data, when you're done with it, the system figures out that you're done with it, so it is automatically reclaimed. But many systems don't do that at all. And it turns out to be an enormous headache for a programmer, determining when a particular piece of memory is free.

When James Gosling [Sun Fellow and vice president, Sun Labs] invented Java technology, he demanded that the language help him take care of this right off the bat. You take the judgment call away from the programmer and give it to the computer.

When the Java software first shipped, it used a garbage collection system that was well known. Over time, we introduced better and better garbage collection technology. There's less performance degradation. We've also focused on improving the performance of large Java programs running on large servers with multiple processors. Some of our customers have acute performance requirements in these settings, and we've been able to develop new garbage collection techniques that meet their needs. So this shows how, even early on, Sun Labs looked at projects in a way that took into account a much bigger picture. Now, as it happened, Java technology became more important and garbage collection took on a prominence we didn't necessarily anticipate.

Q: What is most important to you about the Sun Labs' approach to innovation?

A: Two things. We know that innovation happens everywhere at Sun--not just in the Labs. We are not the only site of innovation. The second thing is we believe process innovation is every bit as important as product innovation. We want to find people who want answers and then go out and solve problems.

Return to the feature.