Contrarian Minds: John Gage

N1 is the third stage of network computing, says Sun's chief researcher.

By Al Riske

10.Jun.04--In a tiny delicatessen in San Francisco's Financial District, John Gage quickly creates and discards metaphors -- musical, chemical, and biological -- for what Sun really does.

This in itself provides the perfect metaphor, because it's exactly what Sun does: We continually change and refine the models we use to design systems.

"We're always striving for simplicity, continuously raising the level of abstraction," Gage says between bites of a turkey sandwich.

Gage has been with Sun since 1982, the year the company was founded, so he tends to take a broad view and think in terms of stages.

In the first stage, he says, we learned to move data among diverse systems. Then, with Java, we started to move bits of software code, and then whole dynamic programs, across the network. Now, with N1, we're moving data center resources and services, which are much more abstract bundles of data and programs, so we can automatically allocate computing power, storage capacity, and network bandwidth.

So, those are the broad brush strokes. But there's much more to the Sun story -- and few are better qualified to tell it than the man Scott McNealy calls "a Sun treasure," the person Bill Joy credits with encapsulating the company's long-term vision in a single, utterly contrarian statement, The Network Is the Computer.

As Gage travels from Bangkok to Paris to Washington, D.C., he acts as Sun's global ambassador. He meets with leaders of government and industry, discusses technology trends and public policy issues, and networks in every sense of the word.

A big part of Gage's job is explaining what Sun does, as clearly as he can, using metaphors and abstractions and oversimplifications. Even the notion of three stages -- moving data, code, and resources -- is an oversimplification, Gage points out. But it's a useful one, because it shows a very real progression in information technology.

The first stage was, at the time, completely contrarian.

"In the 1980s, the accepted wisdom was IBM and the seven dwarfs -- one dominant company and a bunch of niche players with highly specialized technologies," Gage says, "and none of it worked with what the other guy had."

Customers were stuck. They couldn't change. Then Sun came along and pioneered open systems.

"We built an ecology dependent upon communication," Gage explains. "That was our fundamental contrarian idea."

Sun put TCP/IP, now known as the Internet protocol suite, into our very first systems and never looked back.

As it turned out, standardized technology beat specialized technology, hands down. Two things tipped the scales: cross-platform communication and the capability of networked systems to recover from major disasters.

"The network thing, at base, is the realization that you can build, across the network, a kind of reliable system that you cannot build in one place," Gage says.

"In one place, you may have an earthquake or a flood or power outage or some other single point of failure. So Sun began with the insight that, to build reliable systems, you had to look at the network as the computer."

It may seem like a no-brainer today, but at the time, the idea that The Network is the Computer went against the grain. Gage describes how the phrase was coined.

"Bill Joy and I were giving talks in China, and we were trying to think of simple, translatable statements, one sentence long, that captured what we were doing. Bill credits me, but it was a mutual discussion," Gage says.

"The first idea was that computing devices on their own don't matter, but on the network they do. On the Net you have a richness of different capabilities -- floating-point units over here, storage over there, graphics someplace else -- that can all be brought together, and the whole is much greater than the sum of its parts."

The key to making it all work well, Gage says, is the ability to work at ever-higher levels of abstraction, which is exactly what Sun has been doing. Gage describes the evolution this way.

"If I'm designing a chip, I'm modeling the behavior of a transistor. And if I'm good at it, I'm modeling the behavior of clumps of them that do certain computations. If I'm good at that, I'm going to assemble them into certain circuits," he says. "Now that I know that a particular circuit will always add properly, I don't have to design that again.

"There's still the complexity, but you've abstracted it so certain elements vanish. You don't have to concern yourself with them anymore.

"As long as you have defined clearly what one element provides to the others, you can change the makeup of any particular element -- as long as it keeps its promise."

Meaning?

"In terms of data storage, I don't care how you keep this information -- disk or DRAM or flash memory. What I care about is one level of abstraction higher: You keep it, I ask for it, I get it back," Gage says.

"As we solve some problems, we move on to a higher level problem. Today, there are only a few people who build chips, a few who build circuit boards. The higher level problem is bringing various components together into systems."

Java technology has helped us simplify systems because it allows developers to write programs without concern for the underlying system software or hardware architecture. And right from the start, Gage made it his mission to explain this contrarian approach to anyone who would listen.

"He was the key evangelist for Java in the early days and took Java to the far corners of the world," says Sun CEO Scott McNealy, "and places where it's popping up in services and telcos around the world is because he was there first."

Recently, Gage has been talking up the third stage of Sun's development: N1.

"With N1," Gage observes, "Sun is attempting to abstract how the components of the data center -- the software, the storage, the networking, the compute power -- how all those components function, so we can automate change management."

To Gage, N1 is a clear evolution of the network computing idea.

"Sun got it right in its initial view. It focused not on the components but on the network that links them," he says. "Our challenge since then has been to get the abstraction right. But it's never right. It's always an approximation, because things are constantly changing."

So what's next for Sun?

"We began with a technological awareness that everything is on the network," Gage says. "We're now entering the phase of a human organization fully realizing what that means. Distributed communications. Distributed engineering. Distributed functionality. The boundaries of the company eroding.

"There used to be inviolable lines -- there were vendors, suppliers, customers, partners. But on the Net these organizational definitions and metaphors, based on intuitions drawn from experience with containers -- bins, shelves, warehouses, barrels, packets -- change to definitions based on intuitions drawn from flows -- always changing, always interconnected, always evolving feedback mechanisms. Our challenge is to dissolve the organizational structures of our customers, even as we dissolve our own organizational structures, and try to understand how to help them evolve.

"We're not contrarian in the sense of being against something. We're pro-change. We're allowing recombination, reorganization. Dissolving and reforming, dissolving and reforming. We're constantly playing with that."

Gage has the manner of a friendly if somewhat distracted professor, and he is never without his titanium iBook, which he scans for information when a specific detail eludes him.

"All of Sun's work has been focused on expanding the system: downward to the most basic level of tiny peer-to-peer devices all speaking to each other, and upward to these massive, globally connected systems," he says.

"In the past few years, we've been focusing on the systems we build as they merge with the systems that run the major institutions of the world. The human systems. The economic and business systems.

"That's why we're expending so much energy, and the work of a third of our people, to understand: To what use do you put information systems? That's the entire services group, whose job is to understand the business, economic, and societal uses of information."

Sun visionaries are defining the simple rules and simple abstractions that will allow these complex information systems to scale, Gage says.

"Simplicity is hard. You know you're getting closer when you can say one sentence, and your audience says, 'I get it.' They get it because they already understand what you're saying, but in a different context.

"We all know a great deal about complex systems. Complex systems are all around us. Since we're building new ones, we must use the language of those we're familiar with to describe the ones we're creating, which means we need to draw upon biology, natural history, human organizations, music, art -- all of the domains of self-organization -- to communicate, both to our customers and to ourselves, just what we're designing."

Gage pauses for a moment, pushes his plate aside, and looks out the window at the passersby and a lone tree in a giant pot.

"We've moved from the art of the instrument maker to the art of the conductor," he says.

It may not be the perfect metaphor he was searching for when his turkey sandwich arrived, but it'll do quite nicely.