Channel Partners

August 1, 2001

14 Min Read
True Broadband At Last

Posted: 08/2001

True Broadband At Last
Optical Wave-Based Services, Fiber Penetration Spell Upheaval in Local Telecom
By Fred Dawson

Midway through 2001 it appears that optical technology has finally penetrated deeply enough into the local telecom market to ignite the long-awaited broadband revolution.

The spark is leaping across an ever-shorter access distance between end users and fiber connection points, allowing telecom service providers to cheaply marry Ethernet, ATM and other high-speed access modalities with the most cost-effective short-haul transport systems, including Ethernet LAN-grade twisted pair, very high-speed (V) DSL and a wide range of fixed wireless options. Equally important, the expansion of optical wave-based services into metro markets has allowed service providers to put IP and Ethernet to use in ways that upend fundamental cost assumptions of the traditional local transport market.

The impact of this combination is already making itself felt as carriers scramble to set up delivery of gigabit Ethernet services to the enterprise market, long before many issues surrounding use of Ethernet in public transport have been resolved. It all has to do with the fact that, now that corporate bandwidth demand has moved to multiple T1s, DS3 and beyond, external access data rates are coming into parity with the long-standing LAN rates, notes John Curran, vice president for Internet technologies at XO Communications, Inc. (

“The fact of the matter is that when you build a fiber network in a city with 288 pairs and you put optical equipment on to drive those optical rings, the optics side is not all that different no matter what you’re carrying on those rings,” Curran says. “The question is, what do you now put on the customer interface side of that same box? Do you give them four OC-3 connectors or, for the same price, do you give them 10 gigabit Ethernet?

“Because the chipsets are so inexpensive, at the end of the day, the economies and volumes for Ethernet, both fast and gigabit Ethernet, are such that we’re not going to see Sonet services remain as popular as they are today,” Curran adds. “Someone is going to be able to give you twice the bandwidth for half the price.”

The Ethernet-based bargains born of the deeper extension of fiber are not just for the biggest end users, of course. A new class of public network-oriented Ethernet router switches has entered the market, allowing service providers to terminate a gigabit Ethernet link at a fiber node and distribute dedicated 10 or 100 mbps Ethernet feeds over Cat 5 or other low-cost links to customers who are near the node.

“One of the big architectural changes in networks over the past year has been that metro core rings are being complemented by optical access rings in business areas and even residential areas,” notes Frederik Hanell, vice president of marketing for Dynarc (, the Swedish router manufacturer whose U.S. headquarters are in San Jose. “And gigabit Ethernet is a very cheap point-to-point solution if you don’t want to deploy rings.”

The Ethernet opportunity has inspired standards efforts and other initiatives through which vendors and service providers hope to encourage development of interoperable, Ethernet-specific solutions that further drive costs down and make Ethernet over the wide area more appealing to end users. For example, the Resilient Packet Ring Alliance, consisting of over 80 companies and organizations, is close to locking in on a means of creating Sonet-like protection without incurring Sonet overhead. This is a core attribute of Dynarc’s technology, which relies on a technique known as DTM (Dynamic Synchronous Transfer Mode) that is now being made a part of the RPR protocol stack.

The PR media access

control protocol is being standardized within an IEEE working group, known as 802.17 RPR Working Group (, marking the first time the telecom industry is defining a metro network architecture optimized for packet-based services. The MAC defines the method by which multiple packet add-drop devices connecting over a common optical ring share bandwidth in a fair and deterministic manner, meaning the packet traffic moves away from dependence on chance as the arbiter of what gets through at any given moment.

RPR networks will be used as feeder or collector rings in the metropolitan area, doing for packet traffic

what Sonet does for legacy voice traffic, notes Bob Schiff, senior director for strategic marketing at Lantern Communications (, a provider of RPR technology and a leading force behind the alliance. The RPR network aggregates the packet traffic from access points around the ring into highly concentrated packet streams, which are then handed off into the long-haul DWDM and router networks at the network edge.

“It’s amazing how fast, since the first of the year, this (standardization) effort has moved to closure,” Schiff says. “It demonstrates there’s tremendous demand for more efficient distribution of packet services, which is pushing people to make compromises on proprietary technologies rather than engage in the usual drawn-out standards battle.”

By the end of this month the working group expects to begin preparing a draft of the standard and to issue a proposal by year’s end, Schiff says. While a final standard might not be officially approved by early 2003, the concurrence on a draft should lead to commercial implementations of RPR-compliant systems sooner. Already, Schiff notes, Lantern is working with Global Crossing ( in beta testing of the vendor’s packet ring technology as a means for delivering gigabit Ethernet services.

The other new Ethernet-focused initiative is the recently announced Metro Ethernet Forum (, also made up of some of the leading lights in data and telecommunications, which is not trying to set standards but is seeking to identify and promote solutions to key points of inefficiency in the way Ethernet works over public networks. These include interfaces with legacy network management systems, interfaces between Ethernet and MPLS (multiprotocol label switching) and means of optimizing Ethernet for delivery over RPR networks.

The MPLS interface is especially important, insofar as it will provide a seamless means by which IP traffic carried over Ethernet in the metro area can be extended to global reach. This has important implications for IP VPNs (virtual private networks), which can be incorporated easily and at low cost into an Ethernet stream from customers’ premises while providing customers the ability to send integrated packet voice and data traffic over IP networks world wide.

Such activities underscore how suddenly the shifts in optical technology and penetration have made the underlying cost-savings and user-friendly efficiencies of Ethernet an asset for public access networks. But, as important as streamlining the interfaces between Ethernet and other public networking transport protocols might be, it’s clear that many service providers are willing to plunge ahead with Ethernet services without waiting for everything to be optimized to the fullest extent.

Qwest Communications International (, Broadwing Communi-cations (, Williams Communications (, AFN Networks (, Teleglobe ( and XO are among the growing coterie of carriers moving to early deployments of gigabit and, in some cases, 10 gigabit Ethernet services at the metro and access levels. With the spotlight in telecom on failing CLECs, stumbling vendors and bandwidth gluts in certain market segments, those not paying attention to the explosive implications of these carriers’ activities risk being caught unprepared once the Ethernet solution to the pent-up demand for broadband services blossoms into full-scale implementation.

“The demand for services hasn’t slowed down at all,” says Scott Berry, director of global product management at Metromedia Fiber Network, Inc. “People want more bandwidth and more services. They want Ethernet connections. Until the local bottleneck gets solved capacity is going to be unfilled in the long-haul networks”

MFN sells fiber pairs to enterprises and service providers in 27 metro markets along with various levels of expert service, such as provisioning multiple wavelengths and transport management. The “overwhelming application driving demand for our fiber is enterprise LAN extension across the metro region,” Berry says. “Gigabit Ethernet is the wave of the future, riding on WDM.”

Indeed, it is WDM, or more specifically, dense WDM offering multiple streams of dedicated links over a single fiber pair, that has not only made the disruptive potential of broadband over Ethernet a possibility but has made other broadband formats far more cost effective than ever before, notes Mike Friloux, vice president of marketing for AFN Networks, the new carriers’ carrier with over 2,000 route miles of lit fiber now targeted to serve second- and third-tier markets. “We’re deploying as much DWDM as we can in each market,” he says, adding that initial deployments consist of lighting anywhere from ten to 20 waves per fiber.

“Selling optical wave services into the integrated communications environment makes a lot of sense because there’s a certain amount of restoration built into the layer three (IP) and layer two (ATM and Frame Relay) protocols,” Friloux says. “Any one of these gives our customers the ability to play with transport options, using wave service to maximize the advantage of whatever they choose.”

For example, a service provider purchasing wave services from AFN can offer customers protected or unprotected IP connections, he notes. They pay a premium for fully protected links, where Sonet is used to add the fail-safe protection. But they also have the assurance that, using the unprotected option, the layer three protocol is designed to reroute the packet flow on a best-effort basis, thereby providing a considerable amount of protection without incurring Sonet costs. In fact, adds Sanita Krishna, senior manager for product marketing at AFN, customers can choose the unprotected option and than purchase diverse routes for their wavelengths, which, with the addition of a rerouting switch, provides all the protection most service providers are looking for.

The optical revolution has so shifted the economics of telecom business models that it is now cheaper to purchase up to 20 wavelengths than it is to buy dark fiber, notes Stu Verge, executive vice president for network operations at Teleglobe, a Tier 1 ISP that is playing the wholesale and retail sides of the market on a global basis, with special emphasis on selling services to major corporations. “We made a decision that rather than buying dark fiber it made more economic sense, at least in the short term, to buy waves,” he says.

Teleglobe has implemented MPLS over all its links, Verge notes. With MPLS in all the routers over such a network, the carrier increases performance in the IP domain by allowing data to remain within the IP layer of the network as it travels from point to point, rather than moving back and forth between the IP layer and the ATM, Frame Relay or other layers that might be used for transport over various links. At the same time, MPLS allows providers to segregate traffic based on QoS (quality-of-service) parameters so that, for instance, within a given IP VPN (virtual private network) priorities can be assigned to different types of applications thereby assuring that voice or videoconferencing is conducted at toll-quality levels while allowing email and other lower priority applications to flow over the same channel. “IP VPN is our flagship product, with security, guaranteed QoS and burstability,” Verge says.

Teleglobe, with a footpring covering Europe, North America, Asia and Latin America, just launched its IP VPN service along with 10 gigabit Ethernet as a retail wavelength service. The carrier has purchased 10 gigE wave services from Broadwing Com-munications and Williams Communications and is planning to announce a third supplier shortly, Verge says. “In North America we typically are turning up seven or eight (10 gigE) waves per market,” he adds.

The implications of such developments for how voice services as well as data services are positioned and sold are profound. A cost-

driven method of doing things at broadband levels of speed and quality on a par with 10 and 100 megabit-per-second LANs will affect everyone in the channels space, especially insofar as intra-office private-line voice delivered over IP packets in wide area Ethernet connections incurs no PSTN cost add-ons.

One of the carriers preparing to put this proposition to work commercially is Broadwing, which has been operating a voice-over-IP network internally since last October. “About 5,000 employees are using the network today,” says Doug Kellermann, vice president for advanced data services at Broadwing. “We’re able to leverage a hierarchical data network that includes voice on-net to on-net and actually are getting voice performance in excess of PSTN quality.”

Having demonstrated to its satisfaction internally that IP voice over an integrated in-house data network is a viable solution, Broadwing is preparing to add support for voice over IP VPNs as part of its service portfolio, Kellermann says. These are network-delivered VPNs, which means there’s no need for special premises equipment other than standard routers to support them.

“Voice is the biggest value add for IP VPNs,” Kellermann adds. “We have a hierarchical service, so we can offer QoS to support the voice requirements.” The carrier also supports extension of IP VPNs to extranets, with dial-in access from remote locations that allows the VPN to be loaded onto computers in a matter of minutes.

Kellermann is quick to note that IP VPNs and Ethernet are not the only hot products in Broadwing’s portfolio. “The way I see the world is frame relay, ATM and IP VPN have phenomenal potential for us,” he says. “We can offer frame with QoS voice and VoIP with QoS.”

In fact, he adds, the way the market is shaping up it isn’t so much a matter of gigabit Ethernet competing with ATM as it is a matter of gigE sparking a new way of doing things outside the ATM domain. “Ethernet allows you to run the same protocol between sites as you use on site and provision whatever you need seamlessly across that network,” he says. “ATM I see differently, more as a service for call centers where people need to access the Internet as well as voice now that e-commerce is a core part of doing business. Several companies have asked us for bids for that service.”

Corporations who are comfortable with frame relay networks are another factor in the market for ATM, insofar as the voice QoS capability Broadwing adds to frame is really a matter of running frame on ATM, Kellerman says.

As carriers prepare to exploit the new opportunities afforded by the surge in performance and market penetration of fiber, it’s important to remember that whatever happens in the next year or two is but a precursor to the way things will work, once the full potential of what fiber can do to the service mix is reflected in other aspects of networking equipment. In other words, the awkwardness of doing things like Ethernet over public networks today should not be cause for dismissal of the ultimate market potential of these advances.

“One way to think about Ethernet is that there’s a transition here from dollar-per-bit pricing to dollar-per-customer pricing,” advises Andreas Bechtolshiem, vice president and general manager at Cisco Systems ( “In other words, you’re trying to retain your customers and offer them something that makes them loyal to your service, and Ethernet as an access technology looks mighty good in this.”

But Bechtolshiem makes clear that the industry is still a long way from optimizing this opportunity. “What you see today is that the first generation of equipment is enterprise type Ethernet switches that have been adapted or are being extended into this application,” he says. “You would design a system different from the ground up if it had been targeted to the service provider from day one.”

This is especially true of the operations and management aspects, where now, as opposed to the enterprise LAN environment, the SP has multiple customers sharing the infrastructure. “There’s a very important concept here of being able to separate management of the physical plant and what’s going on in the network itself from what’s going on for each of these customers,” he says. “And that’s going to be addressed through OAM enhancements that are still being developed for Ethernet.”

XO’s Curran agrees. “We are effectively five years behind where we should be,” he says. “The equipment at this point is not the type of equipment that you can operate safely a thousand distinct customers with distinct VLANs (virtual LANs).”

But, he adds, “that’s not to say people aren’t going to build and operate Ethernet services over the equipment that’s out there. There’s a brave new world, and some of us will figure out how to scale and operate it.”

Clearly, with the emergence of wave services in the metro and deeper penetration of fiber, market conditions have reached the point where no one can assume that traditional approaches to delivering services will survive for long. The message to SPs of every stripe is: be ready; fiber-based broadband has arrived, and nothing will be the same.

Read more about:

Free Newsletters for the Channel
Register for Your Free Newsletter Now

You May Also Like