Edge-to-core optical switching in the metro network

Th 270893

by Paul Morkel

With the rapid adoption of reconfigurable optical add/drop multiplexers (ROADMs) by wireline carriers, optical switching is being widely deployed across metro core, regional, and long-haul networks. These carriers have begun to realize such benefits as operational simplicity, flexibility of wavelength usage, and a foundation for lower cost-per-bit transport for business and triple-play services�plus scalability to support rapid bandwidth growth driven largely by video applications.

The evolution of optical switching technology now enables new applications for ROADMs in the carrier network. Multidegree ROADM nodes and newly emerging, lower-cost edge ROADM applications can extend the all-optical transmission domain further toward the end user and across the metro core network. A network vision of edge-to-core optical switching is rapidly becoming a reality�with service aggregation at the edge and all-optical, wavelength-switched transport to centralized service creation and content distribution points in the core.Th 270893

Figure 1. ROADM variants have emerged to support requirements in the network edge, core, and hubs.

The state-of-the-art, edge-to-core, all-optical metro network promises key advantages for the carrier, including lower costs driven by a flatter network architecture and new capabilities such as all-optical restoration driven by an integrated ROADM infrastructure.

ROADMs are an overnight sensation more than a decade in the making.

Because they enable high-bandwidth services to be established between optical network endpoints without the expense or delay of truck rolls and manual intervention, ROADMs simplify operations, reduce configuration problems, and streamline service introduction�all of which ultimately drive operational expense (opex) reductions.

Operators of research and education networks have deployed ROADMs to enable cost-effective connections of limited duration between partners in grid-computing and e-science projects. In these applications, the dynamic nature of the ROADM optical layer has supported flexible, time-dependent usage of network resources.

Wireline carriers, on the other hand, have deployed ROADMs primarily for fixed traffic requirements within the transport infrastructure, particularly in support of new services and video applications, rather than the need for dynamic reconfiguration of wavelength connections. ROADM capability, in fact, has now become a mandatory feature in requests for proposals (RFPs) issued by carriers pressured to more affordably and seamlessly accommodate spiraling bandwidth demand and compete for fast-emerging revenue opportunities. Over the last 18 months�triggered by almost simultaneous, breakthrough deployments by multiple Tier 1 telcos�ROADMs have become table stakes for doing business as a contemporary carrier.

Today, the driver for further ROADM penetration in carrier networks is demand for even greater degrees of network flexibility and cost efficiencies. Evidence of this trend can be found on a variety of new frontiers. Demand for multidegree ROADM functionality in the carrier core has increased sharply, and the marketplace has seen the appearance of affordable edge ROADMs that deliver true end-to-end, reconfigurable optical networking. With greater penetration of optical switching, carriers could take advantage of dynamic service provisioning (minutes rather than days) and optical restoration to offer new end-user services.

While ROADMs are now established in carrier ring sub-networks, new multidegree ROADMs are needed to interconnect the rings and ultimately support full, mesh-based optical networking.

Traditional two-degree ROADMs switch wavelengths between a client and network port. Multidegree ROADMs provide switching among four or more network ports, delivering �any-port connectivity� across ROADMs throughout the network topology. In the core, multidegree ROADMs mean no regenerators or back-to-back nodes, so both opex and capex are cut substantially.

The first in-service deployment of commercially available multidegree ROADMs, controlled end-to-end by Generalized Multiprotocol Label Switching (GMPLS), occurred in the U.S.-based Dynamic Resource Allocation over Optical Networks (DRAGON) infrastructure. Interest within the research and education community in multidegree ROADMs indicates a revenue opportunity for carriers, but enterprise customers, too, provide additional motivation. Companies are increasingly interested in commercial grid computing, as well as powerful SAN services and IPTV.

The operational benefits of ROADM technology hold promise for the carrier network�s aggregation layer, as well.

The emerging edge ROADM application arises from changes in network element architecture such as a reduction in the total add/drop capacity at an edge site (for example, supporting 25 percent add/drop of total capacity but still retaining access to any wavelength). If a carrier is willing to accept this tradeoff in the number of wavelengths that can be accessed at a particular site, cost-efficient edge ROADMs create important efficiencies for this segment of the network.

Further simplification can be achieved if the edge ROADM design limits the wavelengths that are accessible at a particular site. Although these restrictions seem counter to the ROADM value proposition, network and traffic patterns are generally simpler and more predictable in the aggregation network; thus, some wavelength-access restrictions are generally acceptable as a cost versus functionality tradeoff.

Key to the edge ROADM value is the seamless optical integration of the aggregation layer with the core network, which obviates regeneration at network tier boundaries. This is particularly beneficial in the case of high-capacity services hubbed to central data-center locations.

Though traditional ROADM architectures offer a valuable level of flexibility in connecting wavelengths, there are some restrictions in terms of where the wavelengths are routed and dropped. For example, a given wavelength generally is connected to a particular fiber port or connector, depending on its color. These fixed, wavelength-specific assignments entail manual fibering at the endpoints of the provisioned service. Furthermore, the add/drop ports are generally associated with a specific network port. Although efficient at reconfiguring add/drop connections at a particular site, current designs typically do not provide a high level of flexibility for client-to-network-port spatial or wavelength switching.

Emergent colorless and quasi-colorless ROADMs remove this operational limitation. Colorless ROADMs allow unrestricted wavelength assignment to fiber ports. This benefit means the carrier could perform wavelength-agnostic preconfiguration and pre-fibering of a transponder, well in advance of when a wavelength must actually be provisioned and a service activated. This ability is particularly beneficial for edge ROADMs with partial add/drop capabilities where access to any wavelength is still required.

In addition, the incorporation of directional switching, meaning the ability to crossconnect any client interface to any network port in a two-degree or multidegree ROADM configuration, provides new flexibility in end-to-end network path establishment and reconfiguration. In addition to greater flexibility for wavelength routing, directional switching also enables optical restoration, including support of alien-wavelength interfaces.

Still in the early days of implementation, ROADM technology could evolve to support a wholly new set of capabilities and flexibility for optical networks (Fig. 1).

With ROADMs more pervasive across the network, optical restoration in the event of fiber cuts, power or device failure, or other disturbances becomes realistically available to the carrier.Th 270894

Figure 2. ROADMs with multidegree switching capabilities at the edge and in the core support optical restoration.

Insufficient restoration times and the inability to guarantee availability of protection paths have historically discouraged carriers from relying on optical restoration. Many of the new services that carrier customers demand today do not require the 50-msec restoration characteristic that is typical of legacy voice services. Nonetheless, carriers strive to maintain this standard, and with increased sophistication in constraint-based routing, we can expect improvements in restoration times, as well as protection-path availability.

In most instances, the GMPLS control plane relies on standardized software protocols to mediate optical restoration. For restoration of externally generated wavelengths (e.g., from an IP router as in Fig. 2), then multidirectional switching is needed in edge and core ROADMs.

In the last couple of years, carriers decided that their networks had to reduce their per-bit cost and make more dynamic, effective use of resources if they were to profitably capitalize on demand for high-speed data services, commercial grid computing, and triple-play consumer services. The fruit of that decision? Intensifying deployment of optical switching via different types of ROADMs from edge to core across the carrier network.

Already, increased ROADM reliance is reducing per-bit transport costs, but the benefits will only magnify over time as ROADM functionality grows more pervasive end-to-end across the carrier network. Carriers are creating automated, demand-responsive, and self-healing infrastructures that will be able to deliver wavelength-on-demand services of unprecedented value. Many of the components necessary for full dynamic reconfigurability�to support time-of-day or day-of-week provisioning for event-driven services such as broadcasts of sports events and other temporary services that require very high network capacity�are being put into place.

In fact, this can be seen as part of a logical, phased development of carrier networks that has taken shape over the last decade. First, we witnessed the convergence of data, voice, and video applications as network operators sought the ease of use, integration, and consistent functionality of a unified network system for optimized Ethernet service delivery over optical networks. The second phase saw accelerated deployment of ROADM functionality, which has enabled today�s more efficient and flexible use of resources. The dawning third phase will bring virtualization of applications and services and fully automated resource provisioning.

With increased deployment of optical switching from edge to core, carriers are readying themselves today for tomorrow�s opportunities.

Paul R. Morkel is director, business management, carrier WDM, with ADVA Optical Networking (www.advaoptical.com).

More in Network Design