Sun harvests research for tech progress

The company's CEO originally opposed the idea of setting up Sun Labs 10 years ago, but the group has created the very products that have allowed the company to keep its independence.

Stephen Shankland principal writer
Stephen Shankland has been a reporter at CNET since 1998 and writes about processors, digital photography, AI, quantum computing, computer science, materials science, supercomputers, drones, browsers, 3D printing, USB, and new computing technology in general. He has a soft spot in his heart for standards groups and I/O interfaces. His first big scoop was about radioactive cat poop.
Expertise processors, semiconductors, web browsers, quantum computing, supercomputers, AI, 3D printing, drones, computer science, physics, programming, materials science, USB, UWB, Android, digital photography, science Credentials
  • I've been covering the technology industry for 24 years and was a science writer for five years before that. I've got deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and other dee
Stephen Shankland
6 min read
SANTA CLARA, Calif.--Sun Microsystems Chief Executive Scott McNealy originally opposed the idea of setting up Sun Labs 10 years ago, but the group has created the very products that have allowed the company to keep its prized independence.

"Scott was against it," said Sun Labs leader Jim Mitchell, a Sun vice president who has been involved with the lab since its creation 10 years ago. McNealy, with an aggressive capitalistic bent, feared the research arm would be "a hole you'd throw money into (without) getting anything out," Mitchell said.

But the labs have promoted another McNealy philosophy: independence. Sun's reliance on inventing its own technology has exposed the company to the risks of isolationism--in particular not being able to tap into the cash cow of Microsoft-Intel computers.

Sun Labs creations, such as 64-bit UltraSparc processors and Java software, paid off in the latter half of the 1990s. Sun's commitment to its Unix server line gave it a huge customer base that competitors IBM, Compaq Computer, Dell Computer and Hewlett-Packard were unable to reach with Windows servers using Intel chips. And Java, which now enjoys widespread use on servers from Sun and others, has attracted developers to the Sun products while undermining some of Microsoft's plans for dominance.

"If you look at the last five or 10 years, they have had the best vision," said Buckingham Research analyst Jay Stevens. "They had the motto 'the network is the computer' long before some companies could spell the word 'network.'"

And while Sun, unlike Dell and others, isn't able to piggyback on research at Intel and Microsoft, it has been able to reserve profits for itself.

"It's probably true that on each PC sold today, Intel and Microsoft make more money than the particular supplier of the product," Stevens said.

Sun's approach has its drawbacks, though, said Technology Business Research analyst Jim Garden. "I think they're terribly practical. Sun overall is quite frugal vs. IBM, which is quite academic," Garden said. "IBM has people crunching out patents, while Sun has people doing applied research to crank out a product to take to market to generate money."

Where Sun's philosophy runs into trouble, though, is during down markets, when not increasing revenue can only come at the expense of competitors. "IBM has over the years achieved a significant inventory of intellectual property. When things get really tough and the market gets really down, I believe it's a real strength for IBM," Garden said. "That's what's required to leapfrog the competition in order to win back market share."

Make or buy
Companies often face the "make or buy" decision when figuring out whether to re-create a competitor's technology or to take the more expedient route of just buying the technology from another company.

Sun, though, fiercely tries to create its own, with Sun Labs directed more than competitors' research efforts at generating products instead of abstract research.

"You don't (buy) other people's technology unless it's a commodity. You should own that intellectual property," Sun Chief Technology Officer Greg Papadopoulos said in an interview Wednesday, as the company showed off Sun Labs technology for the lab's 10th anniversary.

Storage stumble
One area where Sun has learned its lesson in the relation between research and independence is data-storage technology.

"We haven't had a lot of research to help storage," Mitchell said. Instead, Sun tried to acquire storage technology from Encore, MaxStrat, RedCape and others. But with product flops, it had to sign a deal to use storage systems from Hitachi Data Systems.

Research at Sun has been something of a sacred cow, shielded from the layoffs the company announced last week.

Mitchell said the lab budget is based on corporate revenue averaged over the past three years, so researchers are shielded from quarter-to-quarter battles for funding.

Stevens estimated Sun's total research and development at about $500 million per quarter. Sun Labs has about 180 researchers, two-thirds of them working on software and the rest on hardware, Mitchell said.

Sun's research philosophy comes straight from McNealy, whose aggressive business approach infuses the company, most notably in the emphasis on ideas that can become products.

"We do a little bit of basic research, but not much," Mitchell said. "We do what I would characterize as applied research."

Making research pay off can be tough. AT&T Labs, the successor to the famed Bell Labs, is struggling to contribute to the company's revenue. Xerox's Palo Alto Research Center (PARC)--where Mitchell spent 14 years--is famous chiefly for work on technologies such as the graphical user interface, from which Xerox competitors chiefly profited.

Capitalist tendencies also extend to "technology transfer," the handoff of technology from the lab to a product group. Researchers essentially have to sell the technology to product groups, Mitchell said.

"Everyone knows it's hard and mostly doesn't work," Mitchell said. Contact between researchers and product developers is more like rugby than ballroom dancing, he said.

Product research areas at Sun Labs today focus on programming tools, CPUs, getting Java to spread from the smallest devices to the most powerful servers, networked sensors such as thermostats or heart rate monitors, and computers that don't crash.

Sun showed off several demonstrations at the event:

 Programming tools for developers writing software on handheld devices. Where that development today is typically done on another computer that simulates the handheld, Sun's software lets the programmer do the work on the handheld itself.

 Java Badge, a universal identification card that Sun will start using this month. Java Badge includes two small CPUs that handle workplace tasks at Sun such as authorizing a person to enter a building, deducting cash when purchasing food at the company cafeteria and signing on to the company network.

 Awarenex, software that tracks where employees are by which phones or other devices they use. The software lets people control office and mobile phones, e-mail and instant messaging with a single program.

But research doesn't always pan out. Sun still is committed to its Jini software for self-assembling networks of devices or computers, but the software has been a dud so far.

Sun has largely abandoned research on Sparc V10, a next-generation processor design that would overhaul the current Sparc V9 design that's the foundation of the 64-bit UltraSparc line, Mitchell said. Instead, research is pursuing how to put eight or more smaller, simpler chips on a single slice of silicon, he said, an approach that will appeal to Sun's core business customers more than to its less important scientific customers.

Mitchell cites one of his own projects, Spring, as the worst failure. The project was designed to distribute programs to numerous different computers so people wouldn't be reliant on a single, gigantic machine.

The Object Management Group adopted some of the Spring software, and a little bit more was used in Sun's Solaris operating system, but otherwise the technology slipped quietly beneath the waves, Mitchell said.

Prestige, though widely downplayed by corporations, is another motive for corporate research.

High-profile research increases a company's status in academic circles, where the next generation of potential employees is being trained. Indeed, Microsoft measures its research performance by how many academic papers are published.

But judging by the number of recent media events at research groups, corporations are trying to show off to competitors and the rest of the world as well.

HP showed off its labs in June. IBM has let reporters wander the halls of its Almaden, Calif., research facility, peeking at quantum computing and chipmaking equipment. And Sun archrival Microsoft celebrated its research department's 10th anniversary last month.