When Scott McNealy stepped down after 22 years as Sun Microsystems CEO, the attention fell on his successor, Jonathan Schwartz. But to understand how Sun might change in the near future, keep an eye on another executive: Greg Papadopoulos, who took on the new title of executive VP of research and development.
Schwartz put Papadopoulos in charge of a top-to-bottom engineering review of Sun's technology projects over the next three months. Papadopoulos says no Sun engineer will escape his scrutiny, leaving open the chance there may be more changes in the offing than Schwartz's steady-on-the-wheel takeover speeches suggest.
What do all those engineers do?
Sun spends about $2 billion annually on research and development, pouring a greater percentage of revenue into R&D (16.5% in the third quarter) than many of its competitors. But the company needs to shift funds from engineering projects that emphasize the performance of single computers running alone to technology that can boost the performance of a whole network of systems. That's increasingly how IT managers run their business apps. "As we go through and look at the R&D, we say, 'Is this old school or new school?'"says Papadopoulos, Sun's longtime CTO before his title change. "You can't do that informally."
Sun will focus more on designing systems meant to operate in clusters, and it will elevate the importance of software delivered as a Web service, rather than shipped in a box. The ability to do so will define "what it really means to be a computer company" in the next few years, he says.
There are signs that Sun is on the right track. The company recently delivered its new UltraSparc T1 "Niagara" processor, which gains performance by packaging hardware on a single chip that can run 32 simultaneous application threads, each in its own Java Virtual Machine. Those could be search engine requests, Oracle transactions, or other popular workloads. Niagara chips run on about the same amount of electricity as a household light bulb--cooler than Intel or AMD chips in some tests. Sun has been trying to sell T1-based servers to Google, and though no deal has been reached yet, "there's no doubt this stuff is meshed well with the stuff they run," Papadopoulos says. A Niagara 2 version could ship next year.
9,000 Engineers To Assess
Last week Sun also said it would deliver a 128-bit file system for Solaris in June, creating a computing runway for the next 10 years. Sparc and Solaris are still vibrant brands, and Sun's Sparc business actually grew during the third quarter, Papadopoulos says.
Still, customers aren't buying like they once were. Sun's share of the $51.7 billion worldwide server market slipped again last year, to less than 10%. That's its fifth straight year of decline, according to Gartner.
Papadopoulos figures he'll review some 500 projects under way at Sun, with an eye on getting Sun's R&D more in line with customers' desires. "There are 9,000 engineers at the company, and you get asked the question, 'What are they all doing?'" he says. Along the way, he'll identify opportunities for cuts. Will entire projects get axed, or will management use a lighter hand and just redistribute funding among projects? "That's Jonathan's call," Papadopoulos says.
When pressed on how his tenure would differ from McNealy's, Schwartz downplayed any shift in strategy. "The network is the computer," then and now, he said on a conference call. The big vision seems the same. But look a little closer, and it appears Sun is ready to tinker with its formula to try to cure what ails it.