Optical mesh networks really mean business

Network solutions will be focused more on meeting specific carrier business plans than merely deploying cheap capacity.

DAVID SMITH, Corvis Corp.

In contrast to the last few years, when core networks were built largely through venture capital and scientific innovation, the next half-decade will see technology advances driven by specific network requirements needed to make a carrier profitable. A synergistic approach to optical networks-combining the best-of-breed transport technology, management/control systems, and network designs squarely based on an individual carrier's business plan-will help determine which carriers will be successful.

Future of the core
The sheer variety of optical-network solutions in today's marketplace makes it difficult for long-haul carriers to decide which technologies will survive and which will be left behind. "Survivable" solutions can cost-effectively address three major requirements of the core that are expected to peak by mid-decade:

Support new services. More than ever, carriers need to reduce cost structures and roll out new revenue-generating services. Future profits will be based largely on value-added services and less from basic transport services. Flexibility, speed-to-market, reliability, and high performance, combined with the lowest-cost-per-bit transport to deliver these new services, will be key aspects of winning solutions. Equipment choices will be focused more on meeting a specific carrier's business plan than on merely deploying cheap capacity.
Scale profitably. A network must be flexible enough to profitably handle low-capacity traffic levels in its initial phase and high-capacity traffic as a carrier executes its business plan. Although the industry focuses on reducing initial capital expenses associated with scaling a network, there must be a much greater focus on the burden of operating expenses, since these expenses can increase to the point where they eclipse capital expenditures and limit the carrier's ability to be competitive.
Ready for the future. A scaling strategy that constantly replaces old equipment with new equipment as new and better features become available will never be profitable. Flexible upgrade capabilities must be built into network equipment from the start. Also, given the rapid rate of technological advances, it is crucial for a network infrastructure to remain transparent to changes in data protocols, rates, and client services.

No single solution fits all needs. Since real-world business requirements are complex, they may require different solutions in different parts of the network. For example, the carrier may have to combine grooming and aggregation using electronic switches to use wavelengths efficiently, while at the same time deploying ultra-long-haul (ULH) circuits using all-optical switches to minimize the need for regeneration/switching and enable rapid service provisioning.

This need to incorporate multiple technologies into a single solution naturally leads to the concept of layers in the network. The layer concept looks at an optical core as two virtual layers, operating as one network under a single management system (see Figure 1).

Figure 1. A hierarchy of layers enables the core network to scale profitably.

The all-optical transport layer, based on high-capacity mesh architecture, uses optical bypass-allowing traffic to transparently pass through a node without optical-electrical-optical (OEO) conversions-and ULH transport to enable express networking. This layer is optimized for cost-effective support of direct wavelength-based services and high-capacity transport.

The grooming layer packs traffic efficiently into available wavelengths in the all-optical transport layer. Grooming can be performed at the network edge and in the core but for different economic reasons.

All-optical transport layer
The all-optical transport layer is optimized to deliver the lowest-cost-per-bit transport, while the grooming layer is specifically designed for efficiency in the network. This transport layer serves as the foundation for supporting new wavelength-based services that will be the key to revenue generation in the coming decade.

Characterized by all-optical, high-capacity mesh connectivity, the all-optical transport layer supports express networking over a range of distances-over any type of fiber. Although built on a stable, common infrastructure, it must have the agility to simultaneously support relatively predictable traffic demands over long distances and very-high-capacity traffic demands over relatively short distances.

To be future-ready, this layer needs to meet optimum economic metrics over both short- and long-haul transport without fragmenting the network into different localized solutions. Since the layer's underlying infrastructure remains relatively static, ongoing dynamic changes can be accommodated through flexible transmit/receive card options. For example, a carrier can choose to install less expensive cards for circuits running over short distances than the cards required for ULH transport-all without having to make changes to the basic infrastructure. To be profitable, this layer must have the ability to:

Handle multiple-data-rate (2.5-, 10- and 40-Gbit/sec) transport on a common platform.
Scale incrementally, but seamlessly, to support very high network capacities (up to about 15-Tbit/sec total network demands).
Offer rapid circuit provisioning.
Provide service transparency by supporting a range of protocols.
Contain nodes that can combine the ability to add/drop traffic as well as support express traffic bypass-all in the optical domain.

Key to supporting express networking is an all-optical bypass capability, designed to overcome a shortcoming in most of today's optical networks. In conventional long-haul transport networks, 100% of the traffic arriving at each node in the path encounters OEO conversions to add and drop traffic, repack wavelengths, or merely pass through the node.

However, studies of in-service networks indicate that about 80% of traffic in the core is intended to pass through any given node. Forcing all traffic to experience regeneration and termination at every node adds unnecessary expense-as high as 70% of the capital cost of a core network.

Figure 2. Example of an efficient, profitable all-optical transport layer.

On the other hand, an all-optical mesh network, featuring all-optical bypass and ULH transport, enables express traffic to pass through a node transparently, without OEO conversions. That removes a greater part of the electronics in this layer to enhance carrier profitability by lowering costs, supporting latency-sensitive services, assisting fast provisioning, and accommodating higher capacities.

The carrier can gain additional savings for these express spans if the distance between regeneration sites reaches a few thousand kilometers (see Figure 2). With this ultra-long transparency length, intermediate regenerators can be eliminated, in addition to regeneration at the switching nodes.

Grooming layer
The role of grooming is to handle traffic flows in a way that will efficiently fill available channel capacity. This layer performs two types of grooming:

Edge grooming. The "on-ramp" from metro/regional collector networks to the all-optical transport layer manages incoming traffic in subwavelength granularities through aggregation and grooming. Although the access network aggregates the majority of traffic before it reaches the core, grooming of the payload at the edge-down to the STS-1 (52-Mbit/sec) level-can support efficient and cost-effective use of wavelengths across the core.
Selective core grooming. Since efficiency becomes a greater concern in networks with large traffic demands, planners will want to maximize the percentage fill of every wavelength through selective wavelength grooming in the core, since edge grooming alone may not be able to achieve 100% efficient filling of wavelengths.

With selective core grooming, partially filled wavelengths can be diverted off the all-optical-network layer at a small number of selected nodes in the network for re-grooming in the grooming layer (see Figure 3). Even in networks where traffic demands can completely fill one or multiple wavelengths on the majority of connections, these connections will also have some partially filled wavelengths, even when grooming is performed at the edge.

Figure 3. Grooming by business plan, instead of at each node, drives efficiency and reduces operating costs.

This "residual" traffic can consume a large number of wavelengths on a national-scale network. Efficient use of selective grooming in the core can significantly reduce the number of lambdas reserved for this residual traffic, enabling the carrier to use wavelengths with maximum efficiency. Layering the network enables the carrier to efficiently address the problem of optimizing different economic metrics for different functions in the network, without sacrificing the ability to run the network over a single platform.

Combining layers
The two virtual layers physically operate as one network, tied together under a single management system and control plane. While the economics of the core network may be best served by a single-vendor solution, the carrier's business needs extend much further than the core network. Service visibility across the entire network, end-to-end, is a fundamental requirement for any service provider.

Consequently, a management system must not only tie together the layers of the core network, but also provide interoperability with metro and regional networks and legacy equipment. Two industry-wide trends, already underway, help to streamline service activation and operation across the entire network.

The first trend is network-control-system solutions with open, interoperable designs, instead of a historical preference for unilateral custom-build control planes. As this standards-driven trend continues, carriers will find it easier and less expensive to manage and provision a mix of systems.

The second trend combines a network element and operational support system (OSS) with protocols and interfaces that enable the OSS to delegate increasingly more service activation and control responsibilities to network elements. This delegation requires that network-management policies (based on a carrier's business strategy) be implemented at strategic points in the network that provide the most efficient level of control and flexibility.

Bright future
Having an all-optical transport layer built on high-capacity all-optical mesh architecture with fast provisioning and a grooming layer providing edge grooming combined with selective core grooming uniquely addresses the new requirements of the core network. Such a network will:

Support new services. The transport and grooming layers deliver impressive flexibility, speed-to-market, reliability, and high performance to drive down operating costs and support new revenue-generating services.
Scale profitably. A layered core network supports ease of scalability and rapid provisioning to enable the network to expand according to a carrier's business requirements.
Ready for the future. Since all-optical mesh architecture is readily adaptable to higher capacities-and is protocol-, distance-, and bit-rate-independent-this design will continue to gracefully scale to meet even higher levels of network growth and support a more diverse portfolio of revenue-generating services well into the future.

David Smith is vice president of hardware engineering at Corvis Corp. (Columbia, MD). He can be reached via the company's Website, www.corvis.com.