Who Has the Winnng Strategy?
The Battleground: The U.S. telecom market.
The Combatants: The Big 6 network vendors.
The Prize: Leadership in next-generation networks.
Susan O'Keefe and Sam Masud, senior editors
As revenues from the traditional voice infrastructure stagnate and traffic on packet networks soars, the Big 6 equipment vendors--Alcatel, Cisco, Ericsson, Lucent, Nortel Networks and Siemens--are faced with a challenge and an incredible opportunity. Each vendor subscribes to the notion of all-purpose networks, but a battle is brewing over the migration path to those converged
networks. The company or companies with the winning formula could emerge from the fray in the next few years with more than $100 billion in telecom equipment revenues.
Despite predictions that worldwide demand for digital central office switching systems will fall this year, no one expects a wholesale change to packet networking. In fact, analysts predict that in five years, the network will be substantially the same as it is today because of the investment in installed infrastructure. But service providers are expected to increase spending by at least 30 percent over the next few years and those dollars will cement the direction of the industry (See Figure 1). “We are not going to see a revolution in the next five years, but we will know in two to three years what that revolution is going to look like, what technologies are going to win and which vendors are going to win,” said Tom Nolle, president of CIMI Corp. “All the vendors know it and know that if they want to be players in the rest of the world in terms of this 21st century network, they have to be players in the United States first.”
Nolle called the U.S. market the Petri dish in which the 21st century network will grow. Rick Malone, principal analyst with Vertical Systems, agreed. “The United States is not one or two years ahead of the rest of the world; it's four to five years ahead,” Malone said. “The reason the Big 3 in North America are
excited is the same reason the three European players are making aggressive moves into the United States: Demands on the network are causing the need for infrastructure to migrate to higher speeds, provide additional applications and combine voice onto this packet architecture.”
The early moves have been made by the North American players: Nortel bought Bay Networks for its IP expertise; Lucent purchased Ascend Communications for its installed base of ATM switches; and Cisco, the only Big 6 member with a pure data networking background, has made a number of small acquisitions. More recently, Alcatel made back-to-back acquisitions of Xylan and Assured Access, and Siemens made the long-predicted purchase of Argon Networks (a hybrid IP router/ATM switch start-up that has yet to ship a product) and the surprise buy of Castle Networks, a start-up with a product
currently in beta testing that helps bridge the circuit-switched and
packet-switched worlds. Siemens followed up the Castle acquisition with the proposed purchase of IP edge switch maker Redstone Communications.
Although a deal with 3Com was also rumored, it did not come to fruition by the time Siemens announced its new data networking venture, Unisphere Solutions, in March. Ericsson, the leader in mobile communications, thus far has been low key in terms of acquisitions in the United States, although last year it acquired Advanced Computer Communications (ACC), a remote access concentrator maker, and also formed a data networking group last fall. Analysts predict Ericsson will get more aggressive in the coming months.
More buyouts could come fast and furious, creating a rich opportunity for start-ups. Two hot areas every vendor is watching are high-speed routing and services mediation products. Many of the vendors have minority investments in start-ups that provide gigabit/terabit routers such as Juniper, Pluris, Avici,
Nexabit and Netcore, and could seek to buy them outright. Service mediation products from vendors such as Castle Networks, Salix, TransMedia and Sonus are the first generation of so-called agnostic switches and are receiving praise from industry analysts. “These are next-generation Class 4 and Class 5 switches that are really going to have a big play, and the big equipment vendors know that,” said Frank Dzubeck, president of Communications Network Architects (CNA). Malone agreed: “A lot of these boxes are going to be put into carrier networks, especially the ones that can scale, because we're talking about a lot of circuits.”
At a minimum, acquisitions help fill out vendors' product portfolios, although increasingly the big carriers prefer a multivendor environment. So how important is the notion of a one-stop shop? According to vendors, it is crucial because service providers are looking to suppliers not only for products but for systems integration as well. “If your customers are looking to you to do the integration for them, you have a lot more control and ability to do that if you own most of the major components rather than relying on OEM agreements,” said Mike Day, Alcatel USA's director of strategic network planning. “It makes it a lot easier to guarantee service quality and network evolution.” Kevin Oye, Lucent's vice president of strategic and business development for data networking, agreed. “The one thing we don't want is to be viewed by our customers as simply box deliverers,” he said.
According to Dzubeck, even though the product lines of the big vendors are becoming richer and richer, carriers are hesitant to give too much responsibility to one supplier. Still, vendors are getting bonus points for having their own product sets rather than sourcing them from partners. “There are tactical products to provide a tick on an RFP, and those are the products that vendors
can OEM. But a vendor has to own the strategic products and that's why
we're seeing so much merger activity,” Dzubeck said. Siemens is a classic example of a vendor that needed to gain core competencies through acquisitions. It partnered with Newbridge and 3Com for ATM backbone and edge switches and IP service level management; invested in Juniper Networks for backbone routers; and resells 3Com's remote access concentrators. Those arrangements are still in place, but the company gained strength in backbone routing and services mediation through its acquisitions of Argon and Castle.
“Dense wavelength division multiplexing (DWDM), ATM and IP over ATM are going to be the issues that drive these competitive players,” Nolle said. “It's really their positions in those spaces and their strategies for the satisfaction of the voice requirements in this new-generation infrastructure that are going to decide the victories.”
Router Wars
Though none would dispute Cisco's leadership in the router market, there are some analysts who question whether the Cisco 12000 is a true carrier-class product. “Cisco's strength is its customer relationships with the ISPs and its understanding of what needs to be done to build large IP networks, but the 12000 is not a carrier product in terms of being fault-tolerant and NEBS compliant,” said Joseph Skorupa, director of routing and switching at analyst
group RHK. Cisco, analysts said, is rewriting the Cisco IOS from the ground up to make the software modular and portable; it is expected to release the beta version in the second half of this year.
IDC, which separates high-end routers into gigabit and terabit products, does classify the Cisco 12000 in the carrier-class gigabit group along with the Ascend GRF, Lucent PacketStar, Juniper M40, Netcore Everest, Torrent IP 9000 and Nortel Versalar 15000, and gives Cisco praise for its early lead by selling
approximately 1000 of the 12000 routers. “It's Cisco's market to lose,” said Lee Doyle, vice president of data communications research at IDC. Skorupa sees it differently, giving Juniper the nod as the superior performer. “We know that Juniper has over a dozen paying customers including three operational networks with volume deployment. There are three or four start-ups chasing Juniper, but they are at least six months behind.” Cable & Wireless USA became a Juniper customer last month when it agreed to deploy 16 routers in its Internet backbone. Five equipment vendors (Ericsson, Lucent, Nortel and, jointly, Siemens/Newbridge) and UUnet have invested in Juniper and have product distribution rights as well as the right to integrate Juniper's technology into their product lines and services.
The terabit router group is a more rarified club consisting of Avici (in which Nortel has a stake), Nexabit and Pluris. IDC predicts that in 2003 vendors will ship about 6500 gigabit routers representing more than $900 million in revenues; by contrast, only 600 terabit routers will be shipped that same year and will generate about $192 million in revenues (see Table 1).
ATM Thrives
Whether the router or the ATM switch will be at the core of the network is an on-going religious debate within the industry. “Cisco is trying to make that case because they need to; that's their whole line,” said Lucent's Oye. But Junaid Islam, Cisco's group manager for service provider marketing, said that there is room for both IP and ATM. “People have gotten into thinking one network, but there is nothing wrong in having both an IP and an ATM network
if that makes sense. We're focusing on the people who are spending new
dollars, the people spending on IP networks.” CNA's Dzubeck agreed. “It's the newbies [not the mainline facilities-based carriers] who want IP: That's the Qwests, Level 3s and ISPs. They want IP not ATM.”
Despite the focus on IP, Dittberner Associates noted in a recent report that installation of ATM switches worldwide is occurring at a rate that far exceeds the early growth in digital CO equipment. But the firm also warned that the “role of ATM as the dominant switching technology in public networks over the next decade is not assured.” Of the 35,000 ATM systems installed globally, Dittberner gave Cisco the lead, followed by Nortel and Newbridge (see Table
2). Although analysts said that Cisco has done much to revamp its ATM technology since it first sold its ATM switches to AT&T, they give the edge to Ascend's ATM GX550 as a true core switch. As a result of its pending $20-billion acquisition of Ascend, Lucent has bought its way into being the ATM player, with both the GX550 core switch and CBX500 multiservice edge switch in its stable, Skorupa said. Nolle agreed: “Lucent now has an incumbency in the ATM networks of almost every one of the major players. That's worth $20 billion because without it, Lucent would have had to build a
product from scratch and fight to establish it in a marketplace where Cisco is already established because of its deal with AT&T. With the acquisition of Ascend, Lucent has more ATM switches in U.S. service providers than Cisco does.” Lucent's Oye said the Ascend purchase (scheduled to close in May) goes beyond that: “Yes, Ascend has great products today, but what I'm excited about is what we can do with those same people for the next generation of products.”
As for the other players, Nortel is fleshing out its Passport line of ATM switches with the Passport 15000 core switch that is expected to ship this quarter. Alcatel and Siemens have a presence in ATM switching, but Alcatel seems to be backing off in terms of pushing its 1100 HSS ATM switch. Its acquisition of Xylan has more impact in the enterprise market rather than being a strong move in the service provider space. Siemens' MainStreetXpress line
has been developed through a partnership with Newbridge, and one of the products it has contributed to the mix, the 36190, has “not been successful in North America or anywhere else in the world,” Skorupa said. Indeed, last month Siemens/Newbridge announced a five-year deal with Global One, and the initial product in play will be the 36170, which comes from Newbridge.
“Siemens seems like it has been downplaying that relationship with Newbridge a lot these days. With the Argon acquisition, they're buying some equipment of their own, so it will be interesting to watch what will happen there,” said IDC's Esmeralda Silva. Malone anticipated the possible acquisition of companies such as Newbridge or Fore Systems (which is in fourth place with 9.2-percent share of the global ATM market), a view shared by Silva.
“Newbridge would make a decent acquisition for anyone looking to get into that space. They are especially strong in the ATM edge switch area,” she noted.
Although the U.S. is Ericsson's second-largest market after China, the vendor has failed to penetrate this market with its ATM equipment. It did, however, recently win a contract to upgrade Swiss Telecom from a circuit-switched network to an ATM network. With last year's launch of the AXD 301, an ATM switch that scales from 10 Gbps to 160 Gbps, Ericsson is hoping to crack the
U.S. market. “U.S. operators have not deployed these high-capacity switches, but we've got two carriers considering deployment,” said Gary Pinkham, Ericsson's vice president of business development for data networking solutions.
SONET or DWDM?
Despite the emergence of DWDM, a study last month by Communications Industry Researcher cautions that accounts of SONET's demise are premature. As evidence, CIR cites Cisco's participation in the $53-million funding of Cerent, which in February announced a SONET/SDH transport system for aggregating voice, data and video services over SONET running up to 10 Gbps. Lucent's acquisition of Sybarus Technologies, a Canadian SONET/SDH maker, is proof that the “smart money” is still flowing to SONET, CIR said.
Incumbent local exchange carriers (ILECs), which account for more than 70 percent of the SONET market in the U.S., will continue to buy SONET
equipment over the next decade, according to CIR. Market projections from RHK note that, although the SONET market will rise from $4.5 billion last year to slightly over $5 billion in 2002, 10-Gbps SONET equipment will assume an increasingly large share of the overall market.
Although strong in SONET, Nortel, like Lucent, seems to be focused more on DWDM. Siemens, which formed an optical networking group about 18 months ago to focus on the long-haul market, has almost no presence in SONET despite being the No. 2 player in SDH technology worldwide. “Because of our SDH capability, we have technology strength in SONET, but no marketing
strength,” acknowledged Mike McLaughlin, the group's vice president and general manager. “When we looked at entering the market, we saw a lot of difficulty going up against the embedded SONET suppliers. We believe that SONET will be in the networks for years to come, but the majority of growth will move to optical interfaces in gigabit routers and ATM switches. So we decided to enter the long-haul DWDM market because the technology is still evolving so rapidly.” Siemens flagship product is a 32-channel system operating at 10 Gbps per channel.
Cisco does not have a play in DWDM, a market that RHK predicts will grow 70 percent in North America from $1.9 billion last year to $3.2 billion by 2002. But it might make its own move shortly, although it has a cooperative marketing relationship with Ciena Systems, which claims to have an 80-percent share of
the global business for 16-channel systems. “We'll substantially increase our competence in DWDM in the next few months,” said Cisco's Islam, hinting that this could mean a partnership, minority investment or some other alliance.
RHK expects systems with 16 or more channels to be the fastest-growing market segment. Nearly all vendors are either working on or have announced 40-Gbps single-laser optical transmission systems, but don't expect to see any commercial trials before the end of the year. Lucent, for example, recently announced MCI WorldCom-hosted early lab trials of its 40-Gbps TDM-based WaveStar 40G Express and will do commercial testing in Q4 99.
Frost & Sullivan estimated that Lucent has the largest share--30 percent--of the DWDM market in the United States. “We've put 80 wavelengths on a fiber and we'll go higher,” said Rich Gitlin, Lucent's CTO for data networking systems. “We're reducing the risks for service providers, and we're giving them the ability to deliver more and more bandwidth.” Alcatel also claims competency in
DWDM, and according to RHK is No. 3 in North America in integrated SONET and DWDM systems (see Table 3).
Although Ericsson has a complete family of SONET and DWDM systems (in December it was selected by Spain's Telefonica S.A. for an expandable 16-to-32-channel system), it has chosen to focus on the metropolitan DWDM market in the U.S. Last month, Ericsson announced a DWDM system that provides the efficiency of a protected ring and is designed to operate up to 500 km, a distance sufficient to serve several small cities. “With the metro ring, you could do Gigabit Ethernet over DWDM on one channel, packet over SONET on another and perhaps a leased line on another channel,” said Roselyne Genin, vice president of optical networks in Ericsson's Network Operators Group. Leveraging its acquisition of Cambrian Systems last year, Nortel also has introduced a metro DWDM system, called OPTera Metro, that can be used in a point-to-point or a survivable ring configuration. Bell Canada plans to trial the system. Cisco, building on its IP expertise, has added the dynamic packet transport line card to the Cisco 12000 router. The card combines SONET restoration principles with LAN capabilities such as packet prioritization for service providers to support the delivery of voice and video services over IP and virtual private networks.
Given the acquisition climate, two new companies in the optical networking space, Lightera Networks and Sycamore Networks, are also attracting the attention of industry watchers. In fact, Ciena acquired Lightera in mid-March. Both specialize in optical switches that make it easier to provision high-speed private lines such as OC-48 connections. According to Dzubeck, a company such as Sycamore is an example of how fast advances in technology can come. “Sycamore has gone from product inception to shipment in nine months, when in the optical world you're talking about product development taking multiple years,” Dzubeck said. Nor are service providers reluctant to buy products from start-ups, as evidenced by Williams Network's $24-million pact for Sycamore products. Executives at these new companies are often on their second or third start-up and have well-established contacts among service providers.
The Slow Road to Convergence
In the current IP-versus-ATM debate, Skorupa believes that too much is being made of whether IP is a suitable protocol for voice traffic. “I prefer to call it voice over non-circuit switched networks. The first place we'll see the
deployment of packet voice is in trunk networks--that is, replacing the tandem switches with packet switches whether they are IP or ATM. With these switches you've got what's called the tyranny of DS0 because everything has to be converted back to DS0 everywhere in the network. That's where companies like Sonus, Salix, TransMedia and Castle are focused. With both voice over IP and voice over ATM there is too much discussion about voice bits riding for free. But telephony is not about bit transport, it's about services like call processing, billing and so on, and that's where Nortel and Lucent have
great strength.”
All of the vendors recognize that the market demands not boxes, but new solutions, while allowing carriers to preserve their collective trillion-dollar investment in the existing infrastructure. With its recent announcement of the Succession Network, which SBC, AT&T and France Telecom are testing, Nortel offers carriers a way to move telephony services to the multivendor ATM network that they are using today to provide data services, while leaving the door open to migrate the service to an IP network.
Skorupa praised the introduction of the Succession Network, saying this kind of decoupling of solutions demonstrates that companies such as Nortel and Lucent understand that, in the future, they may have to do business with customers that opt for another vendor's hardware. “Nortel has a good story to tell because three or four years ago they took their software and rewrote it from the ground up to make it more modular and portable,” Skorupa said.
Despite the pace of change, Skorupa said that service providers should not feel they have to make a choice between IP and ATM as an all-purpose platform. “Convergence will happen in multiple stages. Some people will build pure ATM networks; others are saying that if they can collapse six networks to two that's better than going to one because now they don't have to force fit everything,” Skorupa said. “So they're building a purely IP network optimized for IP and lighting one color [channel] on the [DWDM] network, and they're building an ATM network for everything else and lighting up another color. We don't believe this myth of convergence that says it has to be a single technology.”
That means the European players still have a chance to get in the game in the United States, Silva said. “Siemens, Ericsson and Alcatel may be behind right now in terms of where Cisco, Lucent and Nortel are, but true convergence hasn't even begun. So it will be a race to see who gets products out and who signs up the accounts.” Lucent's Oye agreed that there is a lot of change ahead. “I could draw a map of different product segments that are supported
today, but I think all of that is going to get thrown up in the air. What protocols are going to be riding over the network of the future and the services that are delivered will see an awful lot of evolution. I don't think there's a lot of people saying that any of the protocols that are around today will absolutely be the protocol five years from now.”
Susan O'Keefe and Sam Masud are senior editors at Telecommunications.
telecommagazine.com |
