October 12, 1999 Web posted at: 11:35 a.m. EDT (1535 GMT) by Elisabeth Horwitt

From...

(IDG) -- Rob Carter, corporate vice president and chief technology officer for FDX Corp. in Memphis, has no illusions that the best network product always wins. "I've seen great technologies go nowhere for market reasons," he laments.

It wasn't always so. In the industry's early days, engineers ran companies, and technical excellence was what counted most to customers. Marketing was an afterthought at best.

This changed, however, as networks grew to become crucial underpinnings for mission-critical applications. End users and upper management got into the IT purchasing loop. So savvy vendors hired marketing experts and launched major ad campaigns to convince nontechnical buyers that a new switch, server or e-mail package was just what they needed.

As a result, the network graveyard is strewn with the corpses of worthy products and technologies that failed, not through technical shortcomings, but because their vendors miscalculated or were simply outgunned in the marketing wars. This overview explains why several solid products met an untimely demise.

MORE COMPUTING INTELLIGENCE
	IDG.net home page
Network World Fusion home page
Free Network World Fusion newsletters
Digital: The revolution overtook it, 1/26/98
Vendors, users rally around 100VG-AnyLAN, 2/6/95
Reviews & in-depth info at IDG.net
		IDG.net's bridges & routers page
		IDG.net's hubs & switches page
		IDG.net's network operating systems page
		IDG.net's network management software page
Year 2000 World
Questions about computers? Let IDG.net's editors help you
Subscribe to IDG.net's free daily newsletter for network experts
Search IDG.net in 12 languages
News Radio
	Fusion audio primers
	Computerworld Minute

1. Lack of marketing

Through the 1980s and into the 1990s, a surprising number of leading vendors maintained the belief that excellence would sell itself, so they never bothered much with marketing or advertising. Banyan was one such company; Digital was another.

"Digital was a company founded and run by engineers whose philosophy was, if the product is good, people will beat a path to your door," says Paul Jorgensen, a former Digital field engineer who is now a network technical specialist at Washoe Health System in Reno, Nev. This belief was one reason why many of Digital's network wares never expanded much beyond the firm's existing customer base.

2. Doesn't fill a need

A related blunder made by marketing-challenged product developers is to assume that customers are as enamored with pure technological sophistication as the developers.

Remember those management frameworks of the late '80s that were supposed to handle the entire computing infrastructure, including the kitchen sink? Like Digital's Polycenter and Nynex's Allink, they've all since disappeared, or as was the case with AT&T's Unicenter and British Telecommunications' Concert, they've been incorporated into a network management outsourcing service.

While many of these products included useful capabilities and sophisticated technology, customers balked at spending hundreds of thousands of dollars and months or years of implementation time on an amorphous architecture with dubious paybacks.

At the time, customer demand was for modular systems such as Hewlett-Packard's OpenView, and for specific tools to address distinct network management needs, says Frank Dzubeck, president of Communications Network Architects, a Washington, D.C., consultancy.

3. Priced too high

A common cause of a product's demise is a consumer's decision that the extra features or superior capabilities offered by the product aren't worth the added price.

"Token ring was a very elegant way of sharing bandwidth on a LAN," recalls David Passmore, president of NetReference, a consultancy in Sterling, Va. "What killed it was LAN switching because all that matters there is price per port. Ethernet was simpler and supported by more vendors, so it was cheaper to build Ethernet chip sets."

The fact that token ring inventor Olof Soderblom demanded a license fee for every chip set sold didn't help the architecture, either.

In the case of ISDN, it was the service providers who balked at the cost. Because it was expensive to install ISDN in the central office, local carriers in particular were slow to deploy the technology, says Gary Habermann, director of technical resources for Widener University in Chester, Pa.

"By the time they did, it wasn't a leading-edge technology anymore, but passé," he says.

These days, digital subscriber line and cable modems are rapidly filling the broadband access market that ISDN targeted. "ISDN won't vanish, its growth will stop dead," says Habermann, whose university still uses ISDN for specialized applications such as videoconferencing.

4. Dueling standards

When multiple vendors, or alliances, simultaneously introduce different technologies that address the same need, the groups often engage in marketing tactics to establish their technology as the de facto industry standard. The loser gets labeled proprietary and fades away. Remember Betamax?

Take 100VG AnyLAN, a 100M bit/sec LAN for twisted-pair wiring that was developed by Hewlett-Packard and initially backed by AT&T Microelectronics and IBM. AnyLAN had good functionality and performance, and wasn't a difficult upgrade from Ethernet, says Fred McClimans, chairman of Current Analysis, a consultancy in Sterling, Va. However, the 100Base-T and 1000Base-T groups touted their work as the next Ethernet standard, although the technology had little to do with Ethernet. Plus, 100Base-T was cheaper.

A much more one-sided standards war pitted Ipsilon against the big switch vendors. Introduced in early 1996, Ipsilon's IP Switch ATM 1600 was heralded as the next network wave, NetReference's Passmore recalls. The device married the speed of switching to the control functions of routing.

Major router vendors were months away from developing a comparable technology. But customers who were closely tied with Bay (now Nortel Networks), Cisco or IBM were willing to wait rather than take a chance on a start-up's proprietary product.

Eventually Cisco rolled out Tag Switching, which became the basis for the widely supported Multi-protocol Label Switching standard. The final blow to Ipsilon's IP switching scheme was the appearance of Application Specific Integrated Circuit-based terabit routing, Passmore says. In late 1997, Nokia bought the company for $120 million - less than 25% of Ipsilon's estimated worth in 1996.

Microsoft is, of course, the undisputed champion when it comes to using its dominance of one market to cut off a competitor's air supply in another. But the titan is hardly alone.

"Microsoft and Netscape gobbled up market share from all the small independent e-mail firms with nice technology and features," Passmore says. "By bundling e-mail into their software, they shut down the marketplace. Look at Qmail. Look at Eudora."

Not all victims of such flank attacks are small and helpless: take IBM, for example. "OS/2 was a superior product that was clearly well ahead of Windows in functionality, management, single logon and drivers, at least through Windows 2.X," Passmore says.

While IBM sought to drive OS/2 purchases from the top down, Microsoft was capturing the desktop market. Windows had an edge because it evolved from DOS. "And as soon as Microsoft had clear dominance with Office, the game was over. It was too hard to get the suite to work on OS/2," he says.

5. Proprietary apron strings

History has shown that major systems vendors don't always make the best network vendors. Too often, they allowed, or even encouraged, the market to perceive their network business as a sideline to the systems business. By refusing to support other vendors' platforms, the companies effectively limited a product's reach to all but their existing customers.

"Sun's SunNet Manager ran only on Solaris and only managed Solaris systems, not Unix," says John McConnell, president of McConnell Associates in Boulder, Colo.

Moreover, the vendor focused on selling its workstations. These decisions are why SunNet Manager lost out to HP OpenView and IBM NetView despite its strengths and eventually moved into the telephone company management niche, he says.

Similarly, "Digital's Vaxcluster and DECnet Phase 4 were great network products for DEC systems," says Jorgensen. "When PCs got popular, there was only one company that made a DECnet stack for PCs." IBM had similar troubles when it tried to pit its Advanced Peer-to-Peer Networking, the nonhost-based version of SNA, against TCP/IP.

Digital took the profits generated by network hardware such as Gigaswitch and fed them into its Alpha product line, McConnell says. As a result, most of the good network engineers left the company, he says.

Survival of the fittest

The death of a product can be frustrating and expensive for network executives who rely upon it. That's why you should research business and market-related factors in addition to technical considerations and cost before you make a major buying decision, says Ed Risinger, president of HealthData Resources, a software manufacturer in Boston.

"Products often fail because the market didn't move their way. That's the American system, and it's beautiful."