Optical Ethernet Comes into Vivid View

by Dana Cooperson and Brian Van Steen

The rise of optical Ethernet brings big business opportunities into vivid view for service providers, but only if their equipment providers see the light and deliver accordingly.

Providers put these business benefits into two main categories: deepened shallow service revenue streams from new and enhanced services; and reduced operational costs, but not at the expense of any core function.

Suppliers that focus their product development, delivery and functionality efforts on these two provider priorities stand to reap the rewards of their efforts. Those that don’t are more than just missing a market opportunity.

So how has Ethernet manifested itself beyond building and campus backbone networks?

Ethernet is appearing in carrier networks as both a transport technology and a service interface. To stay in front in the race to migrate it to an end-to-end WAN technology, service providers will have hard decisions to make about how to lower risk, as they evolve to support scalable, flexible and profitable Ethernet services.

Service providers are adapting existing networks and using a wide range of approaches to deliver low-speed, low-density Ethernet services, but high-speed, high density optical Ethernet services will require new network architectures, equipment and management systems.

Vendors, through the development of new product platforms that are more intelligent, flexible and manageable, can help service providers mitigate some of the risk in evolving their networks and businesses to a more Ethernet-centric world.

Present Priorities
Today, Ethernet services are delivered over many different types of existing network facilities, while Ethernet as a transport technology is employed only for specific links. Without standards for a so-called “Carrier Ethernet” transport infrastructure, core networks remain based on SONET/SDH and WDM, with transport gear adapting to carry Ethernet and other packet traffic more efficiently through added switching and grooming features.

Large networks were built over decades and will:

Evolve from circuit to packet at a pace measured by the need to use the installed assets (e.g., SONET/SDH and ATM equipment);

Continue support for legacy services;

Tie new capital outlays to new revenue sources; and

Provide a foundation for differentiated service offerings.

Service providers need differentiated offerings of high-value services so that they can command a premium market price and return growth to their wireline business. Figure 1 shows the wireline service provider dilemma through IP traffic and revenue statistics over the past five years.

Although traffic growth continues to be strong, each year service providers are receiving less and less per bit carried. From 2000 to 2004, while IP traffic grew 87 percent compounded annually, revenue per bit declined 45 percent.

RHK forecasts that these general trends will continue for at least the next five years. Innovation and intense system vendor competition have helped alleviate the traffic/revenue problem through lower capital costs, but stopping the downward revenue per-bit spiral will take more than cheaper equipment.

Early Ethernet Services
Incumbent service providers selectively launched Ethernet-based services—services with an Ethernet handoff—in the mid-1990s. These services were typically customized offerings to their largest customers. Most of these early deployments were built with point-to-point native Ethernet or ATM switch networks with SONET/SDH transport. Competitive service providers entered the market, and native switched Ethernet approaches increasingly became used by these new entrants as well as the incumbents. Next-generation SONET/SDH systems gained favor as a reliable and efficient means of aggregating Ethernet frames while eliminating the ATM layer.

As Carrier Ethernet evolves, service providers are adopting networks built on MPLS and VPLS cores, which also support Layer 3 services. Meanwhile, Ethernet offerings have evolved from simple point-to-point, best-effort, metropolitan services to include an array of metro and inter-city topologies and features to suit a wide range of enterprise applications and price points.

At this early stage of market evolution, although optical Ethernet offerings provide more bandwidth and easier upgrades than the legacy services they are replacing, this differentiation is insufficient to command as high or higher prices per bit.

In terms of support infrastructure, there are three general approaches to delivering high-speed Ethernet over optical facilities:

Media converters. Simplicity and low cost will continue to encourage service providers to use media converters with dedicated fiber as a last-mile connection option to the first serving CO, particularly as media converters become more intelligent, incorporating service monitoring and test capabilities. However, as scaling becomes a bigger concern, carriers are likely to stop using them for longer-distance, customer-dedicated backhaul.

Customer-located routers. RHK expects this method of interconnect, with a fiber connection directly from the customer premises to the service provider’s existing IP/MPLS network, switched Ethernet network, SONET/SDH network, or ATM network, will continue to be popular, especially for MTUs or other applications where there are efficiency gains from grooming at the customer location. However, improved restoration options that enable services beyond best effort will be more important to support higher service levels.

Next-gen transport equipment. This approach, where packet aggregation and processing power are typically added via Ethernet blades in next-gen SONET/SDH gear, fits with service providers’ requirement to tap the huge investment in SONET/SDH gear, which is carrier class and multi-service.

Optical Ethernet Utopia
Service providers’ search for a simpler, less costly network has motivated system vendors to try to capitalize on the growing importance of Ethernet in the WAN and make real a vision of an Ethernet-based, converged, packet-optical utopia. The future network can be arrived at from different paths, but it is a network that increasingly will require both carrier-class transport features and a high level of packet support.

Many vendors of low-cost, optical, customer-located equipment, like Canoga Perkins, are adding Ethernet transport capability to their fiber extension products with features like loop-back and remote monitoring to improve the manageability of their last-mile products. SONET/SDH transport equipment vendors have evolved their ADM products into next-gen products capable of transporting Ethernet along with legacy services and operating on individual packets in a data stream.

SONET/SDH equipment, with its sophisticated OAM capabilities, is the gold standard in carrier-class networks. Increased cost efficiency in the new products aids the incumbent transport vendors as Ethernet demand pushes them into more direct competition with traditional data networking equipment vendors and start-ups.

Another class of transport vendors, the Ethernet transport device vendors, started with carrier-class, pure packet architectures based around MPLS. These vendors, including Atrica, Corrigent and Luminous, support a move to pure packet networks at the network edge but have widened their addressable markets by adding transport features such as WDM and SONET/SDH PHY and by partnering with larger, more mainstream vendors.

Traditional Ethernet switch vendors, including Cisco and Extreme, have a different starting point and see future networks built around a packet network core with multi-service support at the network edge. Their strategy is to evolve their products to be more carrier class, with IP/MPLS as the bandwidth management and aggregation layer, and SONET/SDH relegated to the transport layer.

The Hybrid Approach
Finally, hybrid approaches will be used by some multi-service switch vendors, including start-ups such as Mangrove and Hammerhead, seeking to take advantage of and help service providers manage the Layer-2 shift from Frame Relay/ATM to Ethernet using SONET/SDH with GFP/VCAT to tie older ATM/Frame Relay networks into a unified IP/MPLS core.

While many approaches to delivering Ethernet service can work when service demand is light, as the number of Ethernet ports grows and service requirements become more stringent, network operators have to consider the overall structure of their networks, their operations processes, and their network management systems to support the growth. The largest service providers will be attracted more and more to evolutionary approaches that offer the resiliency features of SONET/SDH for their Ethernet services.

With the huge installed base of SONET/SDH equipment and the fact that standards to support Ethernet-based networks are still in early stages, SONET/SDH will remain the foundation of optical edge aggregation networks for the current planning period. In fact, SONET/SDH section-level protocols are likely to continue to proliferate in routers and switches as a well-defined, high-reliability, transport protocol layer for IP/MPLS networks.

However, in the network core the high-bandwidth transport fabric to support Ethernet increasingly will be WDM (OTN)-based. Meanwhile, a host of standards bodies—including the IEEE, the ITU, the Metro Ethernet Forum and the IETF—are working to ease the transition between past and future, circuit and packet. 

The authors wish to acknowledge Ovum-RHK colleagues Ron Kline, David Krozier and Mark Seery for their contributions to the research that underpins this article.

Dana Cooperson is optical networks research director at Ovum-RHK.
Brian Van Steen is a senior analyst at Ovum-RHK.