|
|
ONLINE FEATURES
Book Reviews
BW Video
Columnists
Interactive Gallery
Newsletters
Past Covers
Philanthropy
Podcasts
Special Reports
BLOGS
Auto Beat
Bangalore Tigers
Blogspotting
Brand New Day
Byte of the Apple
Economics Unbound
Eye on Asia
Fine On Media
Green Biz
Hot Property
Investing Insights
Management IQ
NEXT: Innovation
NussbaumOnDesign
Tech Beat
Working Parents
TECHNOLOGY
J.D. Power Ratings
Product Reviews
Tech Stats
Wildstrom: Tech Maven
AUTOS
Home Page
Auto Reviews
Classic Cars
Car Care & Safety
Hybrids
INNOVATION
& DESIGN Home Page Architecture Brand Equity Auto Design Game Room SMALLBIZ Smart Answers Success Stories Today's Tip INVESTING Investing: Europe Annual Reports BW 50 S&P Picks & Pans Stock Screeners Free S&P Stock Report SCOREBOARDS Hot Growth 100 Mutual Funds Info Tech 100 S&P 500 B-SCHOOLS Undergrad Programs MBA Blogs MBA Profiles MBA Rankings Who's Hiring Grads | |
SEPTEMBER 24, 2004
By Peter Burrows Azul: A Server Startup with a Plan The odds are long, but this computer maker is set to stake out a major niche by putting the power of complex grids into one small box When he became chief executive of Chestnut Systems two years ago, Stephen DeWitt already knew what Silicon Valley's sharpest minds thought of the company's plan to build a new kind of server computer: Not much. Despite pitching dozens of venture capitalists up and down Sand Hill Road, Chestnut had received exactly one commitment, for a piddly $1 million. "There were lots of jokes about chestnuts roasting on an open fire," says DeWitt. Even trusted mentors questioned his business judgment. When DeWitt asked Motorola (MOT ) CEO Edward Zander, his former boss at Sun Microsystems (SUNW ), to join his board, Zander responded succinctly: "Why does the world need another computer company?" GARGANTUAN TASK. But on Sept. 28, DeWitt hopes to answer that question. The company, renamed Azul Systems, will launch a brazen attempt to shake up the $50 billion server business. The goal: to exploit a major shift in the way software is developed. It's a gargantuan task, but it's off to a good start: Azul has working prototypes of its innovative server in its labs and has lined up a number of top-shelf tech buyers on Wall Street to do field trials this fall. "If it works -- and I have every reason to believe it will -- this could impact the general-purpose server guys," says Motorola's Zander, who has a small investment despite his initial skepticism. What makes Azul's approach different? Today's servers are designed to run a particular brand of software. For example, PCs are tuned to run Windows-compatible programs. But nearly all new corporate software is developed with so-called "virtual-machine" technologies, such as Java or Microsoft's (MSFT ) .net, that let it run on any type of underlying hardware. Azul's server, dubbed the compute appliance, is the first designed from scratch to do one thing: run this "virtual machine" code faster and more efficiently than existing servers. Does the World Need Another Computer Company?As server prices have tumbled in recent years, most experts have assumed that commodity boxes would be all companies would ever need to run their corporate software. Here's why startup Azul Systems might make an impact, anyway:Strong management: CEO Stephen DeWitt is a hotshot entrepreneur with a track record of success. In the late 1990s he created Cobalt Networks, a pioneer of servers based on the popular Linux program, and sold it to Sun Microsystems for $2 billion.Smart money: The startup has raised tens of millions from savvy investors, including venture firm Redpoint Ventures and Ed Zander, the former Sun president who's now CEO of Motorola.Savvy innovation: Azul's first server will leapfrog ahead in the trend from microprocessors that do one thing as fast as possible, to multicore chips designed to tackle many tasks at once. While IBM, Intel, and Sun have plans for chips with 2 to 16 cores, Azul has designed its own chip with 384 cores.But that approach requires expensive software and often lots of consulting services. And it does nothing to control the proliferation of more and more computers to do all of this, which leads to the biggest cost of all: paying IT staffers to install, maintain, and manage all that hardware. Azul's server does the same job, but with one machine. The server plugs into the corporate network, waiting for work if any single machine gets overloaded. That means companies could continue using existing servers and simply offload extra work to a single machine rather than divvy it up into smaller chunks to be handled by several computers across the network. DeWitt isn't just pushing a new product, but a new approach to computing that he calls "network-attach processing." It's akin to what happened in the data-storage business over the past decade. Rather than lock up data in drives enclosed in individual servers, companies began using "network-attach storage" (NAS) setups that created a central pool of disk capacity. That way, each server would never run out of drive space, and the capacity that was available in the NAS could be allocated more efficiently. CORE CONCERNS. Azul wants to do the same thing for processing: create a central reservoir of number-crunching power that can be tapped by any other server at a moment's notice. DeWitt even wants to change the vocabulary of the computer business. Rather than sell a "server," Azul will sell various-size "pools" of processing, delivered through its compute appliance. The product design stands out in two key ways. First is its single-minded focus on running Java-style programs. Other servers spend much of their oomph just running Windows, Solaris, or whatever sprawling general-purpose operating system resides therein. The second innovation is the chip inside the machine. The 105-person Azul has put much of its effort into creating a new kind of processor that's right in line with one of computerdom's latest crazes: multicore chips. Until this year, chipmakers such as Intel (INTC ), IBM, and Sun focused on single-core processors that do one thing as fast as possible. But with further speed gains proving elusive, they've turned to stuffing multiple copies of today's high-end processors onto single slabs of silicon, enabling them to produce multicore chips that handle many tasks more efficiently. MUCH TO PROVE. While Intel talks about "dual-core" chips and Sun plans to move to 16-core varieties by next year, Azul's first server will have up to 384 cores. Its cores will not be nearly as fast as those from Intel and others. But with so many cores, DeWitt thinks Azul's box will be able to process 10 times more software than rival servers. The second-generation box, due out in early 2006, will have up to 896 cores, he says. Of course, DeWitt now has to get from the promising stage to successful execution. The Mountain View (Calif.) company is pitching corporate tech buyers who are still smarting from wasted investments on starry-eyed technologies during the Net boom. Azul, which is only now delivering its first units to customers for evaluation, will have to prove it can deliver huge benefits before they'll place orders after the product ships in volume next year. "Extraordinary claims require extraordinary proof," says Gordon Haff, a senior analyst at researcher Illuminata. Tiny Azul isn't going to threaten the major computer companies -- for now. But since its approach professes to help companies control their biggest budget item -- paying lots of staffers to watch over all those boxes -- many corporate buyers are likely to take a look at the product. Says IDC analyst Vernon Turner: "When something potentially this big comes along, customers will kick the tires." MIGHTY PRECEDENTS. DeWitt knows that the odds for his company are long. "We haven't proven anything yet, and there's no more heavily defended turf in all of the technology market than the server business," he says. Still, he has the opportunity to disprove the notion that hardware innovation is a thing of the past. And Azul has tens of millions of dollars in backing. Success with a hardware breakthrough is far from assured, but it wouldn't be unprecedented. During the height of IBM's mainframe dominance in the late 1960s, companies such as Digital Equipment and Data General exploded onto the scene with minicomputers that even midsize companies could afford. In the early 1980s, those companies were in turn vanquished by newbies such as Apple Computer (AAPL ), Compaq, and Sun. Odds are that Azul will never approach that level of glory. But with a bold new idea and a charismatic front man, it likely won't fade away without fanfare, either.  Burrows is Computer editor for BusinessWeek in Silicon Valley Edited by Beth Belton
BW MALL
SPONSORED LINKS
Buy a link now!Get BusinessWeek directly on your desktop with our RSS feeds.  Add BusinessWeek news to your Web site with our headline feed. Click to buy an e-print or reprint of a BusinessWeek or BusinessWeek Online story or video. To subscribe online to BusinessWeek magazine, please click here. Learn more, go to the BusinessWeekOnline home page  | | |
 Copyright 2004, by The McGraw-Hill Companies Inc. All rights reserved.Terms of Use | Privacy Notice |
Printer-Friendly Version
