Deploying dynamic wavelength services in a SONET/SDH "legacy" environment

In preparing for the future, carriers must protect current revenue streams. That means the differing requirements of voice and data must be accommodated.

By David Frost
Meriton Networks

Looking back, 2001 will be remembered as the year of legacy in telecom. It's not that any enduring legacy was created in 2001, but that we have all been reacquainted with the legacies of the past. Prior to 2001, legacy technology was a bad thing. New service providers, free of the albatross of an installed infrastructure, could architect the perfect platform to quickly capture the vast revenue streams generated by the next-generation broadband services that end-users were demanding. It was widely believed that the installed legacy networks would not be able to adapt quickly to these demands.

Some new service providers have had success in capturing the broadband service revenues using new infrastructures optimized for Internet Protocol (IP) over Gigabit Ethernet. Established service providers understand the impact of these early successes, and are looking to evolve their networks in response to their customer demands. The dilemma they face is how to evolve their existing networks to be broadband ready without discounting their legacy base.

Contrary to the popular belief at the beginning of 2001, one legacy proved to be a valuable asset for the service provider. What is this valuable legacy that has helped to keep some carriers afloat while others floundered? It's not the installed base of SONET/SDH gear, ATM switches, or IP routers; it's the installed revenue base. This revenue base has allowed service providers to service their customers' new requirements using existing assets in a less than optimal manner to deliver broadband services, and use the revenue to aid the evolution of their networks to an optimal configuration. Unfortunately, new carriers find themselves on the other side of this predicament, needing to expand their revenue base on a platform that is highly optimized for broadband IP. Legacy has reminded us that customers need more than just data or just voice; they need both, and these services are very different in their network requirements.

Exploring options

Over time, there have been attempts to converge voice services on IP packets and broadband IP services on ATM and SONET/SDH. While the concept is sound, its implementation is impractical. Established providers have a very large investment in their SONET/SDH platform to support voice and circuit data throughout their network, as well as a separate installed IP and Ethernet infrastructure.

Using a converged platform implies, at best, the redeployment of a significant number of access platforms, but in reality, means a great deal of customer premises equipment will be scrapped. The OC-3 access multiplexers serving the customer premises today are displaced by the multiservice provisioning platform (MSPP). What happens to those multiplexers that are doing the job very well, and more importantly, are paid for? Similarly, the new service provider must now replace the low-cost Ethernet access switch with a multi-service access device to accommodate the voice and circuit data traffic as well.

One possibility is to put Ethernet cards into the SONET/SDH muxes. While this seems to be a logical progression, there are two issues to contend with. First, while Ethernet is the medium of choice for most enterprise data, it is not the only medium that enterprises require. Fibre Channel, ESCON, native video, dark fiber, and specialty protocols for vertical applications all play a role in end-user networks. These other "niche" data protocols often provide very healthy margins when compared to Ethernet transport. Second, broadband data is variable and dynamic in terms of both throughput and destination. Mapping packet-based connectionless traffic onto a hierarchal structured circuit infrastructure is like mixing oil and vinegar; it requires constant agitation for the best mix.

A dynamic network

Today's metro network requires an optical network capable of providing transparent, dynamic, and flexible transport for the range of services it must deliver. Adding WDM to the optical transport infrastructure adds optical capacity. Integrating the flexibility of a wavelength switching fabric transitions the optical transport layer from a series of inter-worked structures to a flexible, dynamic, and intelligent wavelength networking platform for metro services delivery. From a single optical network, service providers can deploy all the optical services they require in their native formats. The result is a network architecture that can respond to the needs and challenges of today, while enabling new high-bandwidth services in the future.

Key to any dynamic wavelength service offering is management and automation of common operational tasks, including setup, maintenance and teardown of optical paths. A network management system or an intelligent control plane such as GMPLS can provide control of the optical transport network.

With a dynamic wavelength network in place, it becomes possible to offer high-bandwidth connectivity on demand for new services to the existing customer base in response to their requirements. One such example is the provisioning a 1.25-Gbit/sec channel for remote tape backup for two hours a day, enabling a single backup of all an enterprise's data to a centralized secure location. The need to courier tapes off-site is eliminated, and the ability to recover data from the backup is accelerated.

The bandwidth required for such a backup would be cost-prohibitive if it were necessary to purchase dark fiber or monthly services to do it. With the customer's legacy voice and data services providing consistent monthly revenues for the carrier, the additional "on-demand" 1.25-Gbit/sec connection provides incremental revenue over the fiber already serving the customer. With an intelligent wavelength transport network, the important legacies of the past can be supported along with the services of today and tomorrow.