As voice and data equipment makers battle for hearts and minds, one thing is clear: Enterprise networks are poised for a radical transformation.
 By Kimberly Patch and Eric Smalley
Network World, 9/28/98
As in the heady days of the early 1990s, when the words "information superhighway" rolled easily off our tongues, and star-crossed lovers Bell Atlantic and TCI pledged their devotion in a marriage never to be consummated, the theme of convergence is being sounded once again.
The air rings with talk of voice-over-data and integrated multimedia networks. The pundits dream of end-to-end IP networks that carry every essential nugget of corporate information in packet form. They foretell the doom of the giant circuit-switched networks that girdle the globe, the voice networks that will be smashed flat by the crashing waves of data traffic.
Whether you agree with the prognostications or believe we may never see the Promised Land of a single network infrastructure, the convergence mania sweeping the network industry anew is affecting you and your enterprise network. That's because these days data network vendors and telecommunications equipment companies, as well as venture capitalists, are behaving as though convergence is a fact of life.
>From product design to acquisition plans, companies such as Cisco, 3Com, Ascend, Nortel and Lucent are basing their strategies and tactics on the common vision of convergence. As a result, two industries that have largely kept to their own hunting grounds in the past are now attacking each other's turf.
The Ciscos and 3Coms of the enterprise world are striking deeper into the carrier infrastructure market. They hope to quickly shore up their voice expertise and the reliability of their products to capitalize on the rich data network buildout taking place among service providers. For their part, the Lucents and Nortels of the telco realm are striving to buck up their data network arsenals and are taking the converged network fight back to the enterprise market.
The dramatic shifts in the network landscape have equipment makers scrambling to redefine themselves as being all things to all people. These shifts also portend fundamental changes in the very nature of enterprise networks. And despite convergence's promise to simplify networking, the reality is likely to be a good deal messier.
Network managers will face a number of tough decisions, including whether to outsource once-private networks, how and where to take advantage of voice-over-packet services and which applications will force the issue of integrating voice and data. And far from being able to turn to any one solution, network managers will have to weigh an even greater number of options, if not vendors, than they do today.
"Convergence doesn't necessarily mean a single technology does everything," says Paul Sagawa, senior research analyst at New York-based investment research firm Sanford C. Bernstein. "It means the coexistence of multiple technologies and the ability to mix and match and make a network according to the real needs of your users. It makes the job of the network manager much more subtle and more difficult."
In some ways, the psychology of convergence could prove to be the biggest stumbling block for network staffs. "Network managers have to resist the tendency to look at the new options through the filter of past experience," Sagawa says. "The future is going to look so different from the past that it's no longer instructive to rely upon experience in making choices. It's going to take being much more flexible in one's approach to networking than it does today."
Pitfalls and promises
One of the more tantalizing promises of convergence is the consolidation of separate network infrastructures, bringing with it cost savings and simpler administration. Ironically, achieving converged networks could mean substantial capital investments.
"The biggest problem in this whole theory of convergence is the notion that you have excess capacity to put calls onto," says Maribel Lopez, a network strategies analyst at Forrester Research, a Cambridge, Mass.-based market research firm. "That's not a valid assumption because most people don't have a bunch of extra bandwidth just waiting around."
Any quality-of-service (QoS) scheme that requires a lot of bandwidth or service levels can't be guaranteed, notes Hans Schwarz, senior vice president of enterprise network systems at Siemens. And voice requires the highest level of service. "Overprovisioning will be necessary," he says.
Rosy cost-saving scenarios for converged voice and data networks can lose their luster when bandwidth requirements are factored in. "You have to try to recapture the cost of the bandwidth and the gear, which means your payback time is much longer than you thought," Forrester's Lopez says.
Also, convergence does not necessarily mean consolidation. "You are still going to have some public switched telephone calls, so you're still running two networks no matter what," Lopez says. The notion of convergence reducing the management burden is at best an oversimplification.
At the same time, the push toward convergence and the competition between the data and voice network vendors will likely benefit enterprise users over time by driving networks to well beyond 99% availability, Siemens' Schwarz says. If data communications vendors want to win in the carrier space, they have to start matching the levels of reliability and network uptime that their telco rivals have had to ensure in the public network. The improvements data communications vendors adopt will make their mark on tomorrow's corporate networks.
Also, old distinctions that once seemed permanent and often dictated purchasing decisions will fade into the background. For example, the carriers' virtual private network (VPN) services are dissolving the LAN-WAN barrier, according to Sanford C. Bernstein's Sagawa. Increasingly, functions such as remote access will migrate from LANs into the WAN, he says. With bandwidth on demand as part of VPN services and the coming of service-level agreements from carriers, network managers will be able to establish end-to-end QoS levels.
What's more, carriers will increasingly become your equipment suppliers as they package LAN/WAN gear along with their managed service offerings. For example, Sprint's highly publicized Integrated On-Demand Network (ION) offerings essentially are managed services that will include customer premises equipment. ION and other services in the coming wave of integrated carrier products could relieve network managers of the burden of selecting, configuring, managing and - better still - upgrading the equipment at the edges of their networks.
Carrying data
While equipment makers of every ilk are paying lots of attention to enterprise networks, it is in the carriers' massive networks that the future of network gear is being forged. Voice accounts for a diminishing portion of the traffic in the nation's telecom infrastructure. Within a few years, by many estimates, data will eclipse voice on the carriers' networks. The carriers' need is clear: convert from voice networks that can carry data to data networks that can carry voice. That is why companies that traditionally supply carriers with high-end switching gear - Lucent, Nortel, Siemens and Alcatel - are scrambling to gain data network technology and experience.
"From a service provider perspective, if for no other reason than eventually I'm going to be carrying more data than voice anyway, I'm going to optimize for a data architecture," says Robert Redford, director of product marketing for Cisco's multiservice switching business unit. "And I'm going to set that up so I can carry different types of services such as voice just as well as I can carry them over a circuit network today."
Carriers might even offer voice services at no charge as an incentive for companies to buy data services. "The amount of revenue and profit from voice is getting smaller and smaller all the time," Redford says. "It's not a value-added service. It's a commodity service. And we all know from our business classes that you can't sustain price on a commodity service."
It is true that voice services are difficult to differentiate and because of this, they will be put under tremendous competitive price pressure in the carrier marketplace, adds Tom Nolle, president of a market research firm in Voorhees, N.J. "Over the next five to 10 years, the margins on voice services are going to decline, and carriers are going to be seeking to earn higher profits from services that have features, which is to say nonvoice services," he says.
But voice won't lose its prominence in the carrier networks, simply because it is the way humans communicate most easily. "Babies today are born with mouths and ears and not RS-232 interfaces," Nolle says.
U.S. carriers rake in some $200 billion in revenue annually, and nonvoice traffic accounts for a relatively modest $30 billion of that, according to CIMI. Even as the numbers shift, voice will remain the dominant traffic form in revenue for a lifetime, Nolle predicts. By 2010, voice services will still account for about 60% of carrier revenue but will yield only about 20% of carrier profits, according to CIMI.
Carriers are being driven to support data in their networks not by an impending flood of data traffic but by the need to maintain profits, Nolle contends: "What we're really struggling with is how do you build an infrastructure that has to be optimized in a traffic sense for one thing and in a profit sense for something else?"
One market or two?
The hype about convergence has led telecom equipment vendors and data network equipment vendors such as Cisco, Ascend, 3Com and Cabletron to be portrayed as direct competitors. Cisco already has a strong position in Internet routers, and the San Jose, Calif.-based data network giant already competes with vendors such as Lucent to sell high-end switches to start-up carriers.
Still, the enterprise and telecom equipment makers remain largely confined to their own market spaces. Enterprise vendors have made few inroads into carrier networks, and the telco suppliers, many of which have offered corporate network gear for years, have little market share to show for their efforts.
But that's changing as market momentum increasingly forces the two camps into conflict and convergence on key technologies.
Case in point: Enterprise companies such as Cisco are moving quickly to embrace Signaling System 7 (SS7), a key technology for call setup and support of advanced network features in carrier networks. "The signaling of circuit-switched networks, specifically SS7, is something that the equipment vendors are trying to enable their packet switches to handle," says Tom Jenkins, a consultant at Boston-based consultancy TeleChoice. SS7 would enable the vendors to marry the power of their data network technologies with the intelligence already found in the voice network.
Existing 5E-class switches, the workhorses of the public switched telephone network, will also be transformed, says Susan Barbier, director of market development at Lucent: "They will be able, for instance, to accept data calls and route them across an SS7 network."
The battle lines between the data communications and telecommunications equipment makers are forming around architectural issues. One skirmish has emerged around whether to carry IP directly over SONET, an approach favored by Cisco, or via ATM over SONET, the method of choice for traditional switch makers.
Cisco's position is that IP traffic management, particularly QoS mechanisms such as the Resource Reservation Protocol, are sufficient to replace those of ATM for most applications, and that running IP directly over SONET removes considerable overhead. The switch makers argue that ATM is necessary to support business-grade voice services.
"A lot of the large carriers still want the flexibility of being able to reconfigure their IP networks quickly, and it's difficult to do that with IP over SONET," Sagawa says. "You can't dynamically reallocate bandwidth. So players such as UUNET WorldCom, MCI WorldCom and Sprint have gone to a hybrid approach where they run IP on top of ATM on top of SONET. It's very wasteful, but it does give them flexibility."
Another point of contention is how to implement Multi-protocol Label Switching (MPLS) in carrier networks - throughout the network or as the interface to an ATM network. Again, the issue boils down to whether an IP-only solution can support convergence of voice, data and video.
MPLS, a proposed standard for boosting performance in private IP networks, is an increasingly important ingredient of carrier VPN offerings. Worldwide equipment sales to enable MPLS-based VPN services could exceed $300 billion over the next eight years, according to CIMI.
Of all the data network manufacturers, Cisco is in the best position to gain a significant share of the carrier equipment market. About 60% of Cisco's business comes from the enterprise market, with the remainder coming from the carrier market, Sagawa says.
But Cisco's position in the carrier market "is significantly less secure than it is in the enterprise," he says. "Cisco's gone from being a very modest player on the strength of its Internet router business to being at the table in any serious discussion of future architecture. But it still needs to turn that seat at the table into larger-scale contracts."
The stakes for the data network vendors are a lot higher than simply grabbing a piece of the big telecom pie. Vendors face a number of problems in the traditional enterprise market.
For one, bruising price competition is already rampant, Sagawa says. "It started in some of the less sophisticated product categories, and it appears to be moving up into the more sophisticated categories. I see no reason for that trend to stop now," he says.
CIMI's Nolle says profit margins in the LAN space are thinning, and the familiar enterprise vendors face new competition from companies such as Compaq, Dell, Intel and Hewlett-Packard, which have large non-network revenue streams to fund their forays into the enterprise.
While the data network vendors hustle to participate in the carrier market, the telecom equipment vendors find themselves in a different bind. They already own the big traditional carrier accounts, but they're not viewed as leaders in data network gear, which the service providers need. Nor are the vendors dominant in the ISP market, where Cisco and other data network companies have established a strong beachhead.
Carriers will certainly continue to buy and upgrade their circuit-switching infrastructures, giving the telecom equipment makers some measure of insulation against the data network storm. An analogy can be found in the mainframe market. Despite the earlier claims of client/server proponents about the death of the mainframe, sales continue - albeit at a much lower pace than in the hotter PC server market.
"It's going to be quite a long time before a phone call from me to my local drugstore is going to be over anything but a circuit-switching network," says Tony Rybczynski, director of strategic marketing in technologies at Nortel.
Nolle concurs: "The equipment that's out there now has an undepreciated value of a quarter of a trillion dollars, and if the carriers replaced it all at once, all of it comes right off the bottom line."
Also, revenue from the circuit switches that carriers continue to buy still dwarfs revenue from packet-switching gear. "Any single regional Bell operating company buys more voice equipment from Lucent or Nortel in one year than [the sum of] Cisco's total revenue," Nolle says.
But that's not to say the telecom equipment vendors can take their time acquiring data-handling expertise. The service provider data buildout is already underway, and traditional customers aren't waiting around for telecom equipment makers to get up to speed. Case in point: Sprint's ION, where Cisco has taken the lead role among equipment suppliers.
What's more, some of the most interesting action in telecom is with the so-called IP pure-play start-ups. Carriers such as Qwest and Level 3 Communications are being funded to the tune of billions, and they are giving large contracts to data networkers such as Cisco and Ascend. The start-up carriers are enticed by the low capital expense of packet-switching equipment, says Lisa Pierce, an analyst at Giga Information Group, a Cambridge, Mass., market research firm.
The start-ups account for a small percentage of carrier equipment sales, but their purchases have an impact beyond their volume. They tend to buy leading-edge products, which means they serve as a proving ground for equipment. In doing so, they will help the data communications vendors sell their products to the more conservative incumbent carriers.
Merge to converge
Given the market forces, it's no surprise that the network industry is in the midst of a mergers and acquisitions frenzy. Nortel acquired Bay Networks in June, and there has been a host of smaller acquisitions this year. These include Cisco's purchase of voice switch maker Summa Four in July and Ascend's August purchase of Stratus Computer, which makes platforms for SS7 software. In addition, there's a growing web of alliances among data and voice equipment makers, such as the relationship between Siemens and 3Com. Many industry observers expect more big deals before year-end.
"The consolidation that we're seeing is not about to slow down," says Farrokh Billimoria, a senior technology analyst at San Francisco-based investment research firm Hambrecht & Quist: "Critical mass and size matter now, and global partnerships are becoming more and more important.
The acquisition trend is also not likely to end even if all the second-tier data network vendors are absorbed, Sagawa notes. "A model has been established in which much of the interesting research and development gets done by start-ups, which then get shepherded into the larger players," he says. "It's a viable model that's been proven by Cisco and others."
Another factor fueling the current acquisition binge is the inflation of stock values. This factor makes it cheaper to buy a company for its technology using stock swaps than to try to build the technology.
Enterprise demands
If the demands of data traffic within the carrier networks are steadily and irresistibly driving the evolution of the network equipment market, enterprise networks are producing dramatic changes as they absorb voice.
Voice traffic requires very reliable networks that introduce little delay. Packet-based data networks have historically fallen short of meeting both requirements. However, nothing inherent in voice traffic makes it unsuitable for packet network architectures.
Packet-switched networks have improved to the point in which even voice over the Internet is a viable option for some business applications. Carrying voice over packet networks, particularly private IP and frame relay networks, is allowing many businesses to reduce carrier toll charges or consolidate voice and data networks.
The lure of cost savings jump-started the market for voice-over-packet products. But the force that will likely transform the technology from niche market opportunity to a basic requirement for enterprise infrastructures is the emergence of voice- and video-enabled applications.
A new class of applications - called teleweb, or click to speak - is beginning to emerge. Teleweb applications add a "call button" to a Web site, typically connected to a call center. As they browse, end users can push the call button to speak to sales or customer service representatives. Click-to-speak features can also be used in extranets for conferencing and collaborative applications.
"The possibilities are huge over the next five years," contends Jim Jacobson, interim chief information officer and director of network services for PC maker Micron Electronics. "Just look at the volume of traffic on the Web, and the quality of the infrastructure is getting better as more and bigger providers get involved in building the Web," he says.
Micron is beta-testing a convergence application from Lucent called Internet Call Center. The software runs on a server that sits between the company's telephone switch and its Internet firewall. "We build trunk groups from the switch to the server, and the server takes IP data - that is, voice packets coming from somebody's PC out on the Internet somewhere - and throws them into our telephone queue."
The software allows prospective customers to speak directly to agents in Micron's call center as they browse through Micron's online catalog. The software also allows Micron's agents to push Web pages to the customers they're speaking with.
"We're planning on doing most of our business over the Web in the next three to five years, so we're putting the building blocks in place to be able to handle that, Jacobson says.
Even with the potential for cost savings and new revenue-generating applications, most companies will hesitate to adopt new technology unless they're compelled to make changes to support existing critical applications. A very compelling force, located in Redmond, Wash., is poised to influence many network decisions.
"To a certain degree, Microsoft has more insight into when converged networks are going to be real than we do because they're writing the APIs that all of the application designers will use," says John Hart, 3Com's chief technical officer. Converged networks will take off when some common applications, such as e-mail, implement call buttons that allow end users to initiate voice calls from within the applications, he says.
"My message to the CIOs is, 'Guys, you can get ready for converged networks or not, but as soon as that call button is in Lotus Notes, you know what your employees are going to do with it,'" Hart says. "And when they hit 'call,' you'd better be ready."
|
