BY JULIE BORT Network World, 4/24/00 On a cool fall morning in 1998, researchers at Lucent's Bell Labs in scenic Murray Hill, N.J., meet with some customers. "Invent a device we can use to route signals through our optical networks without turning them into electricity first," the customers cry. By February 1999, the team has a prototype. During a routine meeting with Lucent Optical Networking Group (ONG) executives, the team demonstrates the prototype, which they call an optical router. Jaws drop.Where integration meets R&D E-commerce for intellectual property ONG puts the optical router on the fast track to production. While the researchers work on their invention, business executives fire up the marketing machine for a product name, marketing plan and price points. At the same time, ONG developmental engineers spec out manufacturing costs and equipment, even though the device itself is not completely defined. By that June, the team demonstrates the device for customers, only to be told that it needs more capacity - and has to be in their hands eight months ahead of Lucent's schedule. The researchers slave in their lab to revamp the optical router to customers' demands while producing a reliable product that can be manufactured cost-effectively. On Nov. 9, ONG announces the device: the WaveStar LambdaRouter, a 10-terabit, all-optical router. The beta units ship in February; the production release is scheduled for July. From concept to product, less than 15 months elapse. This is the new R&D - and what a whirlwind it's creating. Once researchers could have studied an idea in their labs for five to 10 years before a business unit would take the technology and spawn marketable products from it. Today, networking vendors deliver products envisioned by customers, like the LambdaRouter, in less than two. Never has the networking industry moved at such a clip and been so innovative. By all accounts, networking vendors have mastered the new business rules, with 1999 emerging as a breakthrough year for intellectual property. Networking vendors earned more patents than companies in almost any other tech-heavy industry, including aerospace, automotive and pharmaceuticals, according to CHI Research, a Haddon Heights, N.J., consulting group. (See chart.) And data communications vendors earned more patents in 1999 than they did in each of the previous five years. IBM placed first, with an amazing 2,736 patents awarded in 1999, compared with its five-year average of 1,800 patents per year, CHI found. (See chart.) Other industrious networking vendors were Lucent, with 1,137 patents in 1999 vs. a five-year average of 759 per year; Hewlett-Packard, with 850 patents vs. 565; and Sun, with 559 vs. 168. The increase in patents is only one of the ways that networking companies have buffed up their overall R&D health, CHI finds. Most of the companies that make it on the Network World 200 list have mastered the craft of research and development. What's less obvious is that they've overhauled R&D in the process. Whereas once a researcher's job was to make discoveries and publish papers, today success is measured by commercial gains. Under the new R&D, labs are required to solve business problems, not explore technology for its own sake. They must create ever more commercially successful technologies - a trick akin to picking No. 1 songs before a musician records them. "This market is unpredictable. Yet the fundamental needs of the network are pretty clear, such as capacity. Who can get there quickest with the right functionality to satisfy the market will win," explains Bob Martin, a Bell Labs vice president and chief technical officer for Lucent.
That's what Inktomi, the biggest R&D spender among NW200 companies, is counting on. It uses its R&D investment to fuel innovation in portal services, network products and wireless Internet technology, says Richard Pierce, chief operating officer at the San Mateo, Calif., company. Inktomi spent $27 million, or 38% of 1999 revenue, on R&D.
Most important, for the first time at Lucent, Inktomi and elsewhere, users have become integral to the process. Researchers now work on customers' expressed business needs.
 "No longer are most engineers locked in the closet with you throwing raw meat in at them. Of course, you'll still have a few of those. But more and more, they've begun to understand that we're expecting a business viewpoint," explains Richard Parvin, director of network infrastructure engineering for Sabre, in Fort Worth, Texas.
Look closerWith research booming, you may be lulled into thinking you don't need to spend time examining the long-term R&D health of any particular vendor. In fact, the reverse is true. Turning ideas into products is the riskiest part of a vendor's business. The overall economic boom has boosted absolute numbers of technology innovations, says Charles Larson, president of Industrial Research Institute (IRI), a Washington, D.C., association of researchers. If the economy slows - as it surely will - only companies with sound R&D methods will continue to thrive. More practically, if you don't examine R&D health, you could become the unwitting guinea pig for products that are not ready for prime time. Or you could be buying into products that a vendor may not support for the long haul, users say. "When it comes to vendors of network equipment, servers and operating systems, a two- to three-year research focus really is an important consideration for us," asserts Deke Kassabian, technical director of networking at the University of Pennsylvania in Philadelphia and a Network World Test Alliance Partner. If a vendor's development resources are too lean, it could struggle to get new hardware designs finalized and into production or new software into beta, Kassabian has found. "Those might be early signs of underspending on R&D, which in the long term might be signs of a product line going stale," he says. Even network executives who have no desire to be on the bleeding edge find it encouraging if a vendor has a good long-term research strategy. Such is the case for Russ Cherry, vice president of Internet technology at CDNOW, Inc. in Fort Washington, Pa. "I take 'relationship trips' to meet with various folks in research departments, to find out if there's synergy between where we're heading and where they are," Cherry says. Determining R&D health is tricky. R&D budgets don't tell the whole story, warns Robert Buderi, author of Engines of Tomorrow: How the World's Best Companies are Using their Research Labs to Win the Future, due for publication soon. "The difference between the 'R' and the 'D' is the difference between central research and research in business units," Buderi explains. "Typically, 90% of the R&D budget is development - turning an idea into a product, while 10% is research, looking for new ideas." The R&D budget gets even more confused under the new R&D rules. Some companies, such as Nortel Networks and Cisco, don't have central research departments. At Nortel, all internal research is attached to a business unit, says David Mann, vice president of technology with the company's Emerging Business Technology Investments unit. Cisco, in turn, relies heavily on acquisitions and equity investment. For these vendors, determining exactly how much is being spent on long-term technology is nearly impossible from R&D figures published in annual reports. Vendors with central research labs now pass more research to the business units than ever before. Business units are responsible for technologies expected to ship as products in less than two years. Business units also create the next incarnations of existing products.Strength in patentsTallying up patents of a specific company is a place to begin, but only that. Research directors can't always predict which technologies will be big licensing opportunities. So some companies have started to patent more, not necessarily create more. Excessive patenting could also indicate a defensive legal strategy, in which companies document legal rights to the technology. "If someone comes to us and tries to collect a royalty, our lawyers can say, 'Well, we've got a stack of patents, too,' " describes Jim Mitchell, a Sun fellow and recently appointed vice president of Sun Laboratories in Palo Alto. Significantly, number is less important than type. Simply put, patents are either strong or weak. Strong patents are those cited by other patents. Weak patents are those that rely on strong patents. A patent is considered exceptionally strong if it is cited 50 times, says Francis Narin, CHI's president. One example is a patent Sun received in 1996 for a method of extracting data from a markup language file and then generating other files, in different markup languages, containing the extracted data. By 2000, other vendors cited this patent 137 times. Microsoft did so in a patent for caching, Compaq in a patent for an e-mail hot link function, Sabre in a patent for an information aggregation system. Strong patents are an indication of what CHI calls "technological strength," determined by a weighted score. CHI's rating system is based on the number of new patents a company receives, how often the company's patents are cited by others (strong/weak) and the age of the technology upon which the patent is based. Because innovation frequently outpaces the patent process, the latter two criteria help give a truer picture of research health, Narin says. Ironically, a company's ability to influence others only gets it halfway to the goal. To score, the company must be influenced by others' technology as well. Narin calls this "absorptive capacity." If a company has high absorptive capacity, it stays on top of big discoveries and emerging standards, and quickly incorporates these into products. For this reason, CHI watches how often a company cites scientific papers in its patents. Absorptive capacity is critical for another reason: It's the only way to keep brilliant people. Engineers who feel included in the scientific community are likely to stay put. So finds Arvind Krishna, director of Foundations of Internet Research at IBM Watson Research Center in Hawthorne, N.Y. "People like to see if there's an avenue for their technology ideas to get out, like standards or open source or even free downloads. Clearly it's a matter of pride," Krishna says. "We need people within our labs to work with the extended core of technology, as opposed to people who are isolated." In addition to using weighted scores from research companies, you can get a sense for absorptive capacity by noting the authors on standards proposals and other open source documentation. Companies with researchers that routinely co-author papers with those from other companies and universities are demonstrating absorptive capacity. Such is the case at Cisco, Lucent, Microsoft, Nortel and Sun. Another key is why a company makes its development choices. Smart vendors use their research not just to create products, but to change the rules of networking. These are the so-called disruptive technologies, in the words of Nortel's Mann. "If you are into a market to attack a competitor, the classic move is to try to change the rules of the game. You typically try to change the economics of a network by employing new technologies," Mann says.Transfer no moreAnother telling indicator of R&D health is a vendor's method for getting new ideas to market quickly. Traditionally, research and development have been two separate events. Researchers would identify a new technology. When they thought it was ripe for a commercial application, they would "sell" it to their internal customers, the business units. If the business units bought into the technology, they'd turn it into products. This process is commonly called technology transfer. "Technology transfer is perhaps the most terrible pair of words ever invented. It creates and shows a serial and reactive process, and it is exceedingly ineffective," declares Lucent's Martin. The most successful researchers have created new models in which discovery and development occur simultaneously. For instance, Lucent uses a fast-track process for breakthrough discoveries, such as the WaveStar LambdaRouter. It matches central lab engineers with developmental engineers, business managers and marketing people in the business units so that a product can be developed while it is being invented. SunLabs relies on a more traditional technology transfer model but lets the technology creators become its long-term champions, and to move with the technology to the business unit for development. They return to the lab once the technology is turned into product, Mitchell says. Nortel, which has no central research lab for the nearly $2 billion it spends annually on R&D, uses an approach called portfolio management. Mann's group, Emerging Business Technology Investments, has crafted itself like a venture capital firm. Working closely with the business units, the engineers and business people in this group identify commercially promising new technologies, internal or external. If external, the group assists in the acquisition and sees the technology deposited in a business unit for development. In fact, buying research is in itself a sign of healthy R&D by today's new rules. "What I want is best of breed. I don't like it when vendors take a not-invented-here attitude. It shows arrogance," says Parvin, who describes Sabre's network as "bleeding-edge" and frequently relies on newly acquired technology from vendors such as Cisco and Nortel. "When a company buys and buys technology, sometimes it loses key [development] people. This is normally only a problem for a short time," says Parvin, who finds that after the initial acquisition period ends, networking companies support acquired technologies in the same manner as in-house ones. In fact, companies that make heavy acquisitions, such as Cisco and Nortel, have literally turned integration into an art form, he says. If your vendor claims that its product reigns supreme because it was developed completely in-house, the vendor may be sorely lagging in research methods. It might struggle with, ignore or out and out lie about integration. In any case, take such statements as an ill omen on research health.User inputDetermining which products will be big financial winners and which will bomb is like picking winning lottery numbers. But networking vendors have started hedging their bets: They're talking to users. In fact, all major vendors interviewed for this article said they've made it a strategic goal to increase contact between customers and researchers. "The new R&D is customer-driven. We're market-focused. No longer are we designing and building a whole product [hoping to sell it]. Today, we get feedback on customer needs and then do the research," says Derek Messulam, general manager of Convergent Solutions at Telcordia Technologies in Red Bank, N.J. The University of Pennsylvania's Kassabian cautions his peers, however. While he's had the opportunity to interact with and at times influence long-term research at the university's key vendors, Kassabian notes that such communication isn't a common occurrence yet. "The vendors are extremely reluctant to allow this sort of direct contact, since it can be a significant time drain on their R&D resources," Kassabian says. "We understand that, and don't want to get in their way. At the same time, we want the opportunity to have some influence and to let them know what we'll be expecting." Lab directors such as Lucent's Martin advise users to be firm and demand the meeting. In doing so, they'll tap deeper into a vendor's intellectual future than they would through warm-fuzzy business lunches. The message from the new R&D: Ask and you shall receive.
Related links
Clinton to unveil proposal for R&D spending
Send this article to a colleague
|
|
 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Responsible for insuring the safety of your network?
NWFusion offers two FREE security e-mail newsletters to help you keep your enterprise network
secure.
Click here to sign-up.
Click here to sign-up.
Three free and easy ways to bring Network World's in-depth editorial content to your own Web site.
Learn more
Learn more