Last stop: packet networks

The big three vendors travel different paths to reach carriers' ultimate destiny

Despite the hype and urgency surrounding packet-based networks, most carriers are rolling them out slowly. Some service providers have legacy network issues to resolve before they drop new equipment into their networks; others have been waiting for the right equipment - or any equipment that marries carrier-class voice with data capabilities - to become available.

Heeding the call to transition to packet networks, a host of smaller companies has emerged. But leading the pack, through brand recognition if not necessarily product capabilities, are Cisco Systems, Lucent Technologies and Nortel Networks.

Lucent and Nortel approach packet-based networking from similar starting points. Both have a solid history of providing reliable, redundant circuit-switching equipment. In surveys, rankings and analyst analysis, Lucent and Nortel top the lists for high-quality, highly reliable central office circuit switches. Their voice expertise is recognized and admired by carriers, and they are the measure by which the industry judges newer entrants.

Ironically, Cisco is one of these newer entrants (Telephony, May 22, page 66). It has built its business by providing an unmatched knowledge of data networking (and some argue, by pushing forward a few non-standard standards for industrywide acceptance).

The company owns between 60% and 80% of the router market, depending on the analysis cited. It dominates the packet world and has increased its push into the service provider market.

These three companies are heating up the tracks under the express train to convergence. Nortel's Succession and Lucent's 7 R/E product portfolios contain the elements needed to migrate or deploy a packet-based network. Cisco's New World architecture, based on its Open Packet Telephony framework, takes the opposite tack: It melds traditional time division multiplexing (TDM) services to its packet-based products.

There is no stopping the packet proliferation. Voice-over-packet calls are projected to zoom from 2.5 billion minutes in 1999 to more than 114 billion minutes in 2004, according to a study from Parks Associates.

Carriers are cautiously gearing up for the growth. "Customers are beginning to be serious about migrating from circuit to packet," says Diane Herr, vice president of Lucent's 7R/E solutions. "Obviously, next generation carriers are starting to implement new networks based on packet, not circuit [technology]." Even more traditional carriers are starting to transition, she says, albeit more slowly.

The Nortel approach

Succession is Nortel's view of the new packet-based infrastructure. "It is the integration of a number of stand-alone components," says Sam Sigarto, vice president of the Succession Solutions Portfolio for Nortel. "It is our fundamental belief that the customer is looking for a solution, not a box." That solution includes the media gateway and communications servers. (Table 1).

Phone switches originally were built as closed environments. It's not easy to design an open system, says Dan Mangelsdorf, vice president of Succession Solutions marketing for Nortel. Though packet-based networking is all the rage, it could not be taken seriously until the migration paths were established. And because carriers are making an absolute shift in their network design and build plans, they want the systems to be open - to outside developers and other technologies.

"We enabled Succession to allow these services to be built and delivered cost-effectively and easily over the coming 18 months," Mangelsdorf says. "As the name suggests, we're selling a network, not individual elements."

Encouraging third-party development is part of Nortel's program. The company claims to have built an open programmable environment so providers and software developers can build on top of the Succession architecture (Figure 1).

To accomplish this, Nortel is using standard protocols and Java Beans, which will propel development of innovative services for the open infrastructure world.

"We're unleashing the power of hundreds of thousands of programmers," Mangelsdorf says. "One person can create a service that takes advantage of the network resources directory or other directories and common things by dragging and dropping [objects]."

The Lucent view

The R/Evolutionary Networking solution lets Lucent aid in the transition to packet networks.

It includes the 7R/E Packet Driver, which enables Lucent's flagship Class 5 switch, the 5ESS, to support packet-based networking (Table 2).

Service providers that don't own the legacy switch can use the 7R/E Call Feature Server, the basis of which is the company's Softswitch. Enhanced voice features such as 911 and call waiting services are supported on the server (Figure 2).

A key difference between Lucent and Nortel, Herr says, is that "the 5ESS switch has been a distributed switch forever rather than a centrally controlled system." Lucent has distributed its software intelligence over multiple processors, and that experience translates well to the 7R/E. "We solved that problem a long time ago," she says.

Several customers are testing the Lucent solution. "Most [tests] are in the lab stage," Herr says. "We expect to see deployment within this calendar year."

Lucent is encouraging third-party development through its Full Circle program, freeing carriers from a single vendor's timetable for new features. It's a shift for Lucent, which, like Nortel, once developed all the features for its own switches.

But the program is consistent with the concept of disaggregating the switch and separating the call control and media gateway functions, says Mike Decelle, vice president of marketing for Lucent's Open Network Solutions group.

"We see ourselves needing to cultivate third parties," he says. "We want to stake out a leadership position in that [developers' space], rather than being a fast follower."

Lucent recently held a Full Circle developers conference for about 300 attendees - developers and providers both were on hand.

At Lucent's invitation, Nortel and Cisco came to review Lucent's application programming interface, Decelle says. "If we're really going to enable a strong third-party [development] community, it stands to reason that first companies like Cisco and Nortel embrace that," he says.

Cisco: A different track

Cisco argues that the fastest way to migrate from a circuit-switched environment to a packet-based one isn't to follow traditional vendor solutions. The company has entered the service provider space with aplomb but seemingly has had a harder time picking up customers on the migration path - perhaps because of perceptions of unreliability or carriers' loyalties to legacy equipment manufacturers.

Cisco's answer to the migration strategies of Lucent and Nortel is the Open Packet Telephony architecture (Table 3). It is a decidedly different approach. It hinges on the MGX 8260 media gateway, the SC 2200 SS7 dial server and the VSC 3000 Virtual Switch Controller softswitch (Figure 3).

"Cisco is clearly coming from the data network side and has been experiencing rapid growth," says Ned Hooper, director of marketing for Cisco's network software services business unit. "Our focus has been on working with our customers to build networks that enable them to take advantage of the growth in data networking."

The company doesn't manufacture Class 5 switches, which must be re-engineered for a packet-based world. Instead, Cisco encourages providers to create overlay networks by capping their investments and moving forward on the Cisco track.

Greenfield carriers are coming to Cisco in droves, Hooper says. "They don't want to install TDM. We don't have to worry about selling around TDM, but [rather] meeting the need of those carriers to get revenue."

Incumbents are a tougher sell. They can see the revenue opportunities resulting from data and converged network services, but they have more complex networks to deal with and build.

"It's a cap-and-grow approach," Hooper says. "They don't want to invest in more Class 5 switches. We give them the 2200 so they can connect into the SS7 [dial platform] directly. They're investing in the data network, not the voice network."

Cisco's application development program, the Ecosystem, has attracted top-notch developers and carriers, Hooper says. The Ecosystem partnership program allows software vendors to build applications for the Cisco platform.

"It's not just about network expansion but how to build networks to go after the real revenue opportunities," Hooper says. "That's where Cisco excels."

Getting some traction

But creating a new network - using any manufacturer's equipment - leads to logistical realities. Back-office systems such as billing, provisioning and management systems must be integrated. Service can't be disrupted; networks must maintain lifeline reliability. With little background in these areas, Cisco has raised eyebrows from competitors and carriers.

Without overtly stating it, Lucent and Nortel consider each the other's key competitor.

Cisco, in their minds, trails behind, lacking the voice expertise and misjudging service providers' need for highly reliable, lifeline networks.

It is easier for a company with voice experience to offer high-quality voice-over-packet solutions than for a data company to add voice support to its offerings, says Lucent's Herr.

"We have the experience and the ability to bring forward all the voice features over packet much quicker than someone else can," Herr says. "We don't have to continue to reproduce voice services. In packet, we can do that. That enables us to spend our investment making sure we have really effective operations systems."

Nortel's Mangelsdorf agrees, adding that Cisco's greatest challenge is to incorporate voice and its related features without losing carrier confidence. "Cisco cannot lower the reliability or service capability expected of today's [voice] networks," he says.

Bob Taylor, president and CEO of Focal backs up that sentiment. Focal plans to use Nortel's Succession solution.

Cisco must understand that 911 service can't change and that law enforcement officials must be able to gain access to wiretaps and call tracing information, he says. Those are not simple tasks to execute in packet networks. Lucent and Nortel "are delivering a greater level of reliable service and the understanding of a service that never goes down," Taylor says. "They [helped create] the most reliable telephone network in the world. We can't take a step backward."

Because Cisco has purchased telecom vendors, it must overcome additional hurdles, adds Nortel's Sigarto. "Cisco has the approach of a patchwork quilt." The company has acquired several other concerns, and now it faces the challenge of integrating the products, network management and control systems, he says.

"If you replace the functional infrastructure with a new one, you have to offer the customer the value, price, reliability and support of the existing services," Sigarto says. "Many carriers are not willing to sacrifice service at all. Others might be willing to start at a lower level, but no one wants a router with best-effort service that can go down for three hours."

But competition in the circuit-to-packet migration game isn't limited to Lucent, Nortel and Cisco. Smaller companies developing products - such as Sonus Networks and Unisphere Solutions - will likely gain a foothold, particularly in the greenfield provider market.

"Nortel and Lucent are breaking up their existing products into separate systems, so they'll compete with everybody," says Tim Brown, chief technologist of voice architecture for Williams. "You can pick the best-of-breed signaling gateway, switch or media gateway."

Though Williams is working with Sonus and Nortel, newer entrants have a tough row to hoe. They might be more nimble, but they face the same issues that Cisco does as an unproven entity. Lucent and Nortel have mature, proven, reliable products, he says.

"We're evaluating where [other companies] are in this space," Brown says. "But you haven't seen any press releases from us and Cisco."

To be sure, Nortel and Lucent still have work to do. Their existing products have proprietary interfaces that must be standardized, Brown says. The distributed architecture also poses challenges for physical and signaling interfaces. Because nodes physically reside in different locations, the interfaces used in a point of presence must be standardized.

"Nortel and Lucent are decomposing the platforms they have today," Brown says. "It's a little bit of new development, but a lot is based around the existing platforms. How the carrier will benefit is from not having to rip out installed equipment."

Williams is predominantly a Nortel shop and is waiting to test the Succession products in its lab, Brown says. "Today we haven't gone too far down that road. It looks like a great plan to move forward."

Culture shock

And move forward they will. Service providers are ready, Nortel's Sigarto says. "This year is the defining year for the next generation infrastructure. They are ready to make this change."

Larger carriers are understandably slower to implement, but they are conducting lab trials and will turn up service soon, he says. Some smaller providers "are ready to be in service now," he adds.

Not surprisingly, the bottom line is pushing the transformation, says Lucent's Herr. "There are two reasons why customers are going to a packet network. The first is that it saves them operations equipment costs. And the other is it represents a new revenue opportunity with converged applications."

One reason adoption is slow is because many of these products aren't available yet. Lucent and Nortel have yet to release some of the products in their respective portfolios. Others are cobbled together or are still in labs.

Carriers are patient, Herr says, because they are re-evaluating their network design plans, too. As carriers plan to deploy packet-based equipment, they must solve the operations issues. For example, Herr says: "They must determine if they want to interface to all their legacy operations or go to a new set of operations systems. They can take this as an opportunity to change their operations environment."

And that takes time, she adds. "Not all are skilled in doing that. It's been a long time since some long-term service providers have created a new network."

The shift is inevitable and monumental, Sigarto says. A transformation like this comes along once every 50 years or so, he says. He equates the migration to an open infrastructure to the migration from analog to digital networking.

Building these new networks is a cultural shift for vendors and carriers alike. "We are changing 25 years of proprietary [solutions] and technology and introducing new ways of doing it," Sigarto says. "The change is not a simple one."

The trouble with having continually changing products is that following their progress is difficult. Product names change, program structures change and confusion sets in.

For example, Lucent Technologies' Softswitch is also the basis of the 7 R/E Programmable Feature Server. It might be a good migration solution, but it's not the best answer for emerging carriers, says Steve Byars, principal analyst for Current Analysis. He recommends Lucent's PathStar access server over the Softswitch.

But the Softswitch, Byars points out, doesn't work with the PathStar. Instead, it focuses on a tandem bypass solution and global wholesale markets. And it doesn't offer Class 5 switch features. That brings up the question of how unified the 7 R/E platform really is, he says.

"It's not a cohesive strategy under one umbrella framework," Byars says. "Nortel's Succession strategy is a more integrated and comprehensive strategy that revolves around two primary products."

And what about Cisco Systems?

"Cisco has many different initiatives that are ongoing," he says. "We're waiting to see which ones bear fruit."

The one commonality the three companies share is that they are striving for systems that are open enough to encourage outside development of new applications.

"Everybody needs partnerships to provide credibility," Byars says. "There is such pent-up demand for the software and technology and for the many varied services and applications no one company can fulfill that demand."