Though 100 Gbps transport is a relatively new phenomenon of optical networking, the technology space is already acquiring nuances of critical importance to both carrier and enterprise network engineers.

For example, the unique needs of the metropolitan network with regard to 100G transport have recently attracted the attention of the Dense Wavelength Division Multiplexing (DWDM) development community. Different approaches to address those requirements have emerged, such as an innovative card architecture that is designed to deliver a quartet of breakthrough efficiencies that makes it optimal for metro deployments.

100G rapidly gained traction as the state-of-the-art speed for transporting enormous amounts of bandwidth, and it initially saw the optical-networking industry align roughly in concert on the same coherent detection model and modulation schemes for enabling technologies. Now, however, meaningful differences are developing among the various vendor approaches, leaving network engineers with interesting strategic choices in bringing 100G to new areas of their fiber infrastructures. Nowhere are those choices more distinct than in the metro.

A Gap in the Marketplace

After 10G transport achieved mainstream status in high-bandwidth enterprise and carrier networking, it took some time, trials and errors for industry to re-orient around the next consensus stepping stone in speed. 40G, for example, was marked by false starts, slow maturation and a limited ecosystem of enabling systems.

During this period of uncertainty, the growth in multiprotocol bandwidth demand on optical networks has never abated. And, now, 100G appears to have simply slammed shut whatever small window of opportunity that 40G experienced.

To fill the need for spectrally efficient long-haul 100G transport, the entire DWDM development community seemed to be working from roughly the same master plan. Coherent receivers, low-noise amplification technology and increased non-liner optical impairment tolerance characterized the wave of 100G solutions targeted first for long-haul transport.

Soon, 100G demand in the metro began to crystalize. After all, it is in the metro that optical networkings most critical, bandwidth-intensive applicationsresidential broadband, business Ethernet, mobile backhaul, data-center connectivity, for exampleall converge. The particularities of this network segment in relation to long haul became apparent, too.

Enterprise data centers in crowded metro areas, where real estate, rack space and power are limited, have faced relentlessly mounting pressure to wring more capacity out of existing equipment and leased connections for higher- and higher-speed traffic, such as 8G Fibre Channel storage area network (SAN) and 10G Ethernet local area network (LAN) applications. Service providers, meanwhile, have started to seek solutions to connect 100G-enabled Internet Protocol (IP) routers simply, via dedicated 100G metro links. The modulation format and coherent detection scheme that were recommended by the Optical Internetworking Forum (OIF) deliver great reach and spectral efficiency. But for common metro applications such as these, the coherent detection technology was over engineered and insufficiently delivered against the various requirements for efficient shorter reach transport.

These drivers illuminated the market opportunity for lower-cost, direct detection 100G designs for the metro, and, indeed, solutions have quickly emerged. What had suddenly disappeared, however, was the industry consensus.

This means that enterprise data managers and service providers must closely evaluate the tradeoffs and engineering choices that have been made along the way in bringing to fruition 100G solutions targeted for metropolitan networks. Non-optimized approaches will draw on too much horsepower and cost too much to achieve the ongoing, long-term cost points that are required in the metro.

Important Differences, Interrelated Efficiencies

A customer-driven, metro-optimized 100G card solution has emerged that uniquely leverages four 28 Gbps DWDM signals to achieve the capacity target. Other designs employ ten 10 Gbps signals (10x10 Gbps) to deliver 100G. The 4x28 Gbps card approach conveys four sets of differentiating efficiencies:

* Cost Initial cost for 100G metro transport via the 4x28 Gbps card architecture can be halved compared to 100G coherent detection designs, and its competitive with 10x10 Gbps approaches that fail to convey the same range of efficiencies.

* Space With the 4x28 Gbps architecture, 100G metro transport can be achieved with only two card slots and five rack units. Thats 50 percent or less of the space demanded by other approaches to 100G transport.

* Power As energy costs continue their rise, the need to reduce operational expenditures has established itself as a business imperative across enterprises and carriers. Fully loaded implementations of the 4x28 Gbps card architecture demand less than 100 watts of power (or, less than 1 watt per Gigabit). 100G transports power demands are cut by half in the metro-optimized approach.

* Spectral The 10x10 Gbps architectures for 100G typically utilize 50GHz channel spacing consuming a total of 500 GHz of spectrum; whereas, the metro-optimized, 4x28 Gbps approach demands a total of only 200GHz of spectrum per 100G line capacity. This amounts to an improvement in spectral efficiency of 2.5 times. The impact is huge. Using the single 4x28 Gbps card for metro-optimized 100G transport, network engineers can realize 250 percent more capacity from existing fiber infrastructurecompared to non-optimized 10x10Gbps approaches.

Every aspect of the metro-optimized, 100G solution was designed to maximize efficiencies, and that singular focus sets the 4x28Gbpscard clearly apart.

Weighing Priorities, Choosing Wisely

Optical networking is not a one-size-fits-all, commodity technology space. For example, a 100G DWDM solution that is truly tailored for the metropolitan network would not prove to be the most appropriate for long-haul deployments. The 4x28 Gbps card that delivers such compelling cost, power, space and spectral efficiencies compared to 10x10 Gbps configurations for metro applications does not reach the same degree of spectral efficiency that 100G coherent solutions provide and long-haul networks demand.

In broad strokes, if the network engineering priority is reach, then 100G with coherent detection is the proper choice. In metro applications to be carried over fiber paths of less than 500 kilometers, in which issues such as space and power efficiency might factor more heavily, the emergent 4x28 Gbps card architecture offers a finely tailored option.

For enterprise data-center managers and service providers, this means that they must carefully review potential equipment partners: Does this vendor offer the breadth of solutions that is necessary to achieve proper tradeoffs and make best use of our end-to-end fiber infrastructure in transporting enormous amounts of data? Can we leverage existing WDM-transport systems? Or are we incurring hard-to-detect costs over the long term by investing in a jack-of-all-trades approach that fails to afford us the flexibility to optimize for both the long haul and metro?

Conclusion

100G is in rapid fashion making its way across more and more of the network, as enterprises and carriers seek an extended break from bandwidth exhaustion and traffic bottlenecks. As the state-of-the-art technology has reached the metro, it has quickly spawned a more diverse landscape of solutions from which to choose. Breakthrough efficienciescost, power, space and spectralare available to those enterprise and carrier network engineers who fully survey the new terrain.