NGN Ventures: Emerging Optical Architectures

DWDM systems need to be simplified and made less expensive, according to Kevin Ranking, President & CEO Tropic Networks, who spoke at this week's NGN Ventures Conference. First generation DWDM has attributes that pigeon hole it into special situations. A major problem is the complexity of provisioning. Every move/add/change in the network requires a tremendous amount of engineering, resulting in a "Launch and Pray" method of provisioning wavelengths. DWDM needs to be closely integrated with existing SONET/SDH systems to deliver an intelligent phontonic layer that's easy to manage and a streamlined network architecture which drives down CapEx. Ranking believes that DWDM systems should connect to existing SONET equipment via ITU-grid optical signals, and they must be compatible with SONET OSS functionality, allowing end to end per wavelength path visibility from the NOC. To further reduce the capital expense of DWDM systems, OEO conversions and the required transponders must be cut and the signal should be kept in the optical domain both inside and between network elements. Tropic Networks offers a metro core DWDM/OADM platform with ability to manage signals on a per-wavelength basis.


Jagdeep Singh, President & CEO of Infinera agrees that the problem with optical networks are their high cost and complexity. As optical networks increase in capacity, the number of required components and the complexity and cost of the network increase as well. This has resulted in the creation of redundant networks and elements to address the needs of each network span: long-haul, ultra long-haul, and metro. Singh also observed that photons are great for transport, but problematic for other aspects of networking. An all-optical network is an analog network, making it impossible to manipulate or measure bits during transport. Other ways to process information are needed. Infinera's has not yet unveiled its technology, but Singh said that its solution will simplify the optical network, enhance service provider portability, and offer managed bandwidth transport.


Dr. John Bowers, CTO of Calient Networks believes that service providers need to migrate from complicated ring architectures to mesh networks while having the flexibility to offer new services as needed without re-building their networks. He suggests that the answer is a transparent all optical network enabled with 3D MEMs switches. Such a network would support real time distributed computing, geographically distributed, large bandwidth networks, and traffic patterns that vary widely over time. Bowers said that 3D MEMs switches provide other advantages, including support of GMPLS provisioning and restoration, transparent switching, and integration of DWDM systems. Bowers predicts that carriers will deploy MEMs switches in order to offer new services and can then take advantage of the operating expense advantages that the technology offers in the network core.


Enterprises already own LANs and SANs, and it is time that they own their entire networks, said Pawan Jaggi, CEO of Celion Networks. Today, WANs are complex and costly, and WAN equipment is designed only for service providers. As a result, enterprises practice "bandwidth avoidance" outside of their LANs. To be convinced to own their own networks, enterprises need to see a quick ROI vs. leased circuits, turn key system solutions, and "touch-free" LAN-like automated operations. Celion's optical solution for enterprises transports signals 6000 km without regeneration from a microwave size box and provides 10 GbE, GbE, OC-192 and OC-48 interface options. Celion aims to provide transport solutions that are distance insensitive and offer a negligible cost of incremental bandwidth.


Simplification of DWDM networks is very much needed, but vendors are not truly dealing with the carriers' problems, according to Dennis Morgen, Network Architecture, Operations & Strategic Planning VP at AT&T. For example, there is still a lot of talk about transport and switching as separate things. These aspects of networking should be merged and viewed in common. Instead of doing TDM grooming, AT&T wants to see it take place at the packet layer. This isn't being addressed enough by the industry. Morgen would like vendors to provide a transformation to a new network architecture. A "pasted-in overlay" will not ultimately provide the needed solution, while a radically new architecture that allows service providers to meet the needs of their customers will be embraced. There are too many suppliers in the optical segment of the industry and the paradigm is wrong. Instead of each vendor trying to sell its own box in isolation, Morgen wishes that these individual box innovations could be combined as a full network solution. This requires companies to merge.


Disruptive, rather than purely incremental innovations are needed to escape the industry's problems, said William Stensrud, General Partner at Enterprise Partners. He warned that the telecom industry is increasingly becoming like the airline industry. Pasting new solutions on top of underlying problems will not result in a competitive, financially sound industry. To be a surviving vendor, a company needs a low burn rate so that it will still be in business when the big opportunity arises, and must be a "scavenger" that opportunistically solves current problems in the mean time. A start-up must make itself indispensable to its prospects, which is the only way it can win business from a service provider.