Journey to the centre of the core
The current business situation is playing a critical role in influencing the direction of core-network development. As bandwidth continues to grow and the overcapacity gap in the core network closes, carriers and service providers are evaluating next-generation core optical-networking equipment with the aim of creating a new optical core.
Not only must service providers realise significant reductions in capital and operational expenses (capex/opex) from this new core network, but also the speed of turning up new optical channels must be significantly improved. In addition, improvements in the differentiated offering of revenue-generating optical services must be addressed.
A pragmatic roadmap has to evolve from today's core optical network—characterised as point-to-point DWDM links with optical-electrical-optical (OEO) regeneration and OEO switching at junction nodes—to the new optical core that not only includes optical transparency, but also full dynamic connectivity. Given that this transformation will not happen overnight, it is important that the appropriate product capabilities and architecture are incorporated right from the start to ensure a credible migration path. It is likewise important that the roadmap incorporates the correct sequence of capabilities so problems are addressed in the right order as dictated by current realities and future requirements.
As the telecommunications industry continues to progress through its consolidation phase, the number one priority that carriers and service providers currently face is to increase their profitability by reducing capex and opex while increasing revenues. Long gone are the days of introducing "cool" technologies just to emphasise brand differentiation to the marketplace.
Also, given the abundance of long-haul capital already deployed, very few, if any, new network-wide builds can be expected in the next couple of years. However, some congested routes where current systems are reaching their maximum capacity will need to be upgraded. Since new DWDM transport systems now cost 60-80% less to deploy and maintain than their legacy counterparts, the decision to deploy next-generation gear on those select routes will be quite evident. This new equipment must therefore prove itself on a route-by-route basis, and not require a whole network for the economics to look attractive. The key will then be to expand these select routes incrementally, in an "in-service" manner, into a multiroute network while continuing to reduce costs on a year-over-year basis.
An increasingly recognised capex metric is cost per gigabit per second per kilometer (euros/Gbit/sec*km). Legacy long-haul DWDM systems just a couple of years ago cost well above EUR30/Gbit/sec*km to carriers. The costs associated with turning up and maintaining these systems were and continue to be (as most of these systems are still being filled) even higher.
For carriers and service providers to seriously consider new systems, capex and opex costs each must be well below the EUR10/Gbit/sec*km level. That must be achieved on a route-by-route basis but with all the right "hooks" in place for an in-service migration to multiple routes and lower costs down the line.
The best approach to achieving that is with optically transparent systems where costly OEO conversions are moved from the network's core to its edges. Not only will this result in significant capex and opex savings, but also incremental channels can be turned up much more rapidly, resulting in substantial customer acquisition/retention as well as revenue improvements for carriers and service providers.
Figure 1 illustrates the relative cost differences between legacy DWDM systems and optically transparent core networking. With legacy systems, connection costs increase with distance, since OEO regeneration (and/or switching) is required every 500 km or so. In contrast, with optically transparent systems, the OEO portion of each transmission remains the same, since OEO conversions are onlyrequired where wavelengths need to be terminated—in other words, at the edges of the core network.
Although the concept of optically transparent networking has been around for some time, too many constraints have prevented any serious deployments to date. These constraints have included engineering complexity, optical performance, operational deficiencies, and economics.
The following capabilities are required for the realisation of optically transparent networking:
- All-reach transport. Optimally addressing the full range (hundreds to thousands of kilometers) of traffic mix on a single platform.
- Fibre-agnostic operation. Operating transparently over different fibre types without the need for OEO demarcation points.
- Dynamic transmission control. Automatically adjusting optical power levels and dispersion compensation due to transient conditions or when adding new channels.
- Multidegree optical bypass and add/drop. Routing any wavelengths to/from any direction without the need for OEO conversions, or simply originating/terminating any wavelength to/from any direction.
- Integrated optical test and performance monitoring. Integrating key optical diagnostic capabilities to simplify the turn-up of systems and incremental channels as well as appropriately monitoring the quality level of each channel.
Additional capabilities such as integrated 10- and 40-Gbit/sec transport should be built-in from day one to achieve further cost savings by appropriately leveraging the economics associated with higher-speed technologies. What is not required, however, at least initially, is photonic switching. Although this capability can automate optical connections and is essential when full agility is required, multidegree optical-add/drop-multiplexer nodes do not necessarily imply that all connections need to be dynamic. In fact, supporting mixed static and dynamic connections would allow carriers to cost-optimise the network design while allowing an elegant migration path toward full agility.
Although fully agile "A-Z" optically transparent core networks will create tremendous value in terms of lowering capex and opex while generating new revenue streams for carriers and service providers, their evolution from today's point-to-point infrastructure is far from trivial. Clearly, there are different paths to get there. What's important for system vendors and service providers is to prioritise the right sequence of capabilities that will gradually build the new optical core and deliver real value along the way.
Before fully dynamic optical connectivity can be realised, DWDM transport issues must first be resolved. Not only is that a definite technical prerequisite, but also it makes complete economic sense in the face of current market realities. By initially focusing on the right transport architecture and capabilities, select route overlays can be deployed in the most cost-effective fashion. Dynamic connectivity can then be incrementally introduced not only when it makes sense (both in terms of technological maturity and economic value), but also where it makes sense. Not all optical connections will initially be required to be dynamic, since this attribute will be largely service-driven. Hence, this gradual static-to-dynamic migration will require core optical-networking platforms to support mixed static/dynamic connectivity, so agile wavelengths can be rolled in without disrupting existing static connections.
Figure 2 offers a high-level illustration of how the optical core network has evolved to where it is today, along with different options leading to the new optical core. The major increment following SDH/SONET rings has been the deployment of point-to-point DWDM systems. From there, the next key evolutionary point was the replacement of multiple back-to-back terminals with OEO switching.
That is basically where we stand today: point-to-point DWDM systems (with typical optical reaches of about 500 km) with OEO switching of all channels at junction nodes. From here, there are two basic paths to the new optical core (A-Z optically transparent core networking).
One evolutionary path is to further augment the switching dimension with photonic switching/user-network interface/Generalised MPLS signaling and the transport dimension to support a fully agile optical network. Not only is this path challenging to deploy, but it is also costly on a route-by-route basis due to all the agility functions that are built-in from the onset.
The other path is to first augment the transport dimension with the right optical performance and, with the right architectural hooks for an in-service migration, later augment the switching dimension with photonic-crossconnect functionality when and where it makes sense in the network. With this approach, interim route-by-route overlays can be addressed in the most cost-effective manner.
The realisation of the new optical core will evolve based on current business realities. Core optical-network solutions will have to address the critical short-term survival tactics of the carriers today with the built-in flexibility to address their long-term differentiation strategies for the future. The pragmatic roadmap to the new optical core requires the right capabilities in the required sequence for today's challenges and tomorrow's growth.
Benoit Fleury, vice president, product management and marketing, at Ceyba Inc. (Ottawa, ON, Canada), can be reached via the company's Website, www.ceyba.com.