Rx for SONET and Ethernet

By LEE JENKINS, Tellabs--Caught between tight capital budgets and shrinking core revenues, today's service providers face a seemingly impossible conundrum: They must figure out how to capitalize on demand for new data services in a profitable way, without disrupting existing networks and revenue streams.

Aug 12th, 2003
Th 129909

The Ethernet market is too large to ignore, but carriers' existing SONET networks are too inflexible for typical data rates.

By LEE JENKINS, Tellabs--Caught between tight capital budgets and shrinking core revenues, today's service providers face a seemingly impossible conundrum: They must figure out how to capitalize on demand for new data services in a profitable way, without disrupting existing networks and revenue streams.

The obvious answer to the data services issue is Ethernet: The market for various flavors of Ethernet-based offerings is estimated at $9 billion over the next four years, according to market researcher RHK (South San Francisco). But Ethernet-only networks have until now only paved the road to bankruptcy court. At the same time, existing SONET transport networks, designed to carry legacy T1 (1.554-Mbit/sec) and T3 (44.736-Mbit/sec) services, are proving too inflexible to cost-effectively deliver 10- and 100-Mbit/sec Ethernet services.

There is a technological resolution to this dilemma, which will become commercially available over the next several months. For those service providers with the foresight and resolve to move ahead, Ethernet-over-SONET (EoS) technology, with virtual concatenation (VC), will allow carriers to cost-effectively upgrade their networks.

Technology can't provide all the answers, however. Service providers must address key business issues, including how to avoid cannibalizing existing T1 and T3 service revenues and how to update operational support systems (OSSs) to automate handling of these new services in a manner that enables profitability in the near-term.


Service providers have acknowledged the growing interest in Ethernet services within MANs to provide virtual private networks or LAN extensions. Given Ethernet's domination of the LAN market, a public-network service that connects multiple LANs without the complexity of converting or encapsulating the Ethernet traffic into a different data format has obvious appeal for service providers and business customers. Such services are available today from local service providers - including competitive carriers, many of whom are struggling financially - and on a limited basis from incumbent service providers. The incumbents typically use traditional Ethernet switches over dark fiber networks or overlay an Ethernet service onto existing transport networks. These approaches can meet the immediate needs of business customers on a special-case basis but are unsuitable for a widely available tariffed service offering.

EoS offers the opportunity to develop a mass-market Ethernet data service, which is very similar to the highly popular frame relay (FR) services launched in the early 1990s and virtually ubiquitous today. Similarities include flexible bandwidth assignments, service-level-agreement (SLA) guarantees, and active performance monitoring and reporting. All are value-added services demanded by today's business customers. The major difference between FR and EoS is that FR is typically a sub-T1-level service and EoS is a multimegabit service. More important, there is a way for service providers to develop these new services, which won't require a network overhaul or the addition of yet another layer of technology that must be separately and expensively managed.

In today's networks, Ethernet and particularly Gigabit Ethernet (GbE) services deployed on a mass-market basis would quickly chew up massive amounts of available metro bandwidth. An OC-48 pipe, operating at 2.5 Gbits/sec, could only support two GbE services, with the remaining 500 Mbits/sec of capacity stranded.

EoS and VC allow service providers to carve up Ethernet traffic into T1-sized or STS-N-sized chunks, which can then be mixed with other T1 private line traffic and transported over SONET. By building on the existing transport networks, service providers avoid the expense of a major network overhaul.

Fundamental to this approach is a shift in deployment of Ethernet-capable technology from the network edge using optical edge devices to the core with existing SONET transport gear. Where newer SONET add/drop multiplexers (ADMs) have been deployed, this step is as simple as adding a line card to an existing SONET ADM. In most major metropolitan areas, however, where demand for Ethernet services is greatest, the SONET transport gear is five years old or more and can't support Ethernet services that simply. In these situations, it is possible to deploy an optical transport switch to "interwork" with existing digital crossconnects and ADMs and deliver Ethernet and GbE services directly over the SONET transport network.

Although the market for this equipment is just emerging, equipment vendors are already aware that they must design these new transport boxes to be high in function and relatively low in cost. To make it possible for service providers to add another element to the central office, vendors must avoid bundling in functionality that can drive up the price of this new gear on a first-cost and an ongoing operational basis. One way to do this is to take advantage of existing transport equipment; don't build-in unnecessary functions that could prove difficult to manage and make the overall solution too expensive for mass-market use.

One function that should be built-in from the outset, however, is the ability to natively handle additional Layer 2 packet switching. The thought here is to futureproof the network against the eventual day when all traffic will be packet-based, including voice services. While the telecommunications industry slowdown has prompted many industry observers to revise their forecasts on how quickly the circuit-to-packet transformation takes place, there is little doubt that it will happen, and it is better to be prepared for that eventuality in this network upgrade. Initially, this new transport switch can be interconnected to other network elements over ring-based, high-speed optics, but ultimately the EoS function should be integrated into existing systems like digital crossconnects via an in-service upgrade.

Operating EoS in legacy networks

New network technology is of little use without the OSSs. Telcordia Technologies, which developed 80% of the OSSs in use in North America today, has already issued a draft specification on how EoS-VC will fit into existing provisioning and network management systems. The first specifications release for Telcordia's OSMINE process is due out by the end of the first quarter; OSMINE helps enable network equipment compatibility and interoperability with Telcordia OSSs, so that flow-through provisioning of services without extensive manual intervention is possible.

Automating the provisioning, alarms, performance monitoring, billing, etc., of Ethernet services is an essential step in making these services available on a mass-market basis. No doubt adding Ethernet services to the SONET transport network ratchets up the level of complexity. Instead of T1s and OC-Ns to manage, there will be multiple variants of virtually concatenated channels to address. While this flexibility is crucial to enabling services that match the needs of business customers, it must be handled with the appropriate tools.

On the current timetable, equipment vendors could go through the OSMINE process by the middle of this year, allowing service providers to proceed on course with their deployment of next-generation SONET to enable Ethernet services. Companies not using Telcordia OSSs -new local service entrants, surviving competitive local-exchange carriers, and independent telephone companies - will have an easier time adapting their OSSs to manage Ethernet and GbE services. These providers can use the element management systems developed by equipment vendors to get started or integrate that functionality into their existing OSSs. Long-distance companies that capitalized so vigorously on FR services in the early 1990s could discover a new, highly lucrative market at a time when revenues from other services aren't growing, and may even be shrinking due to price competition.

Get the tariff right

The new flexibility that EoS with VC will bring has one other consequence for local service providers: It opens up a Pandora's box of tariff possibilities, which have to be carefully considered.

Adding new services is a fruitless endeavor if they cannibalize existing revenue streams such as the solid base of T1 and T3 services carriers sell to business customers today. It is possible to avoid such disruption by establishing Ethernet tariffs in the intermediate range between these services. There is a window of opportunity at the 6-35-Mbit/sec range that will satisfy businesses constrained at one to three T1 lines but uninterested in or unwilling to pay for the higher-priced T3 services (see Figure).

The most logical approach is to establish a set of tiered services in 5-Mbit/sec or more increments. That gives customers the widest range of possibilities, without destroying the T1 revenue base, which is already under fire due to local service competition and associated price pressures.

The challenges of offering Ethernet services, including operational issues and the delicate matter of tariffing, pale in comparison to the opportunity that awaits service providers. With EoS technology and the proper business case, service providers can chart a course to a more profitable and growth-driven future.

Lee Jenkins is director of product-line management for Tellabs, Metro Networking Group (Naperville, IL). He can be reached atlee.jenkins@tellabs.com.

Ethernet becomes more economical than T1 at about three to four T1s worth of capacity (or 4.8-6.4 Mbits/sec). DS-3 crosses over with Ethernet again at about 22 T1s worth of capacity (or 35 Mbits/sec).
More in Earnings Statements