Metro 100G and the coherent brick wall
By Dr. Maxim Kuschnerov
The future never works out quite how you planned. After introducing 100G coherent technology into core and submarine networks, product design engineers now lust for a progression of proprietary coherent interfaces for the metro sector, enabled by falling component costs and higher-order modulation schemes that decrease the cost per bit of the physical interfaces.
These higher-order modulation schemes include 200G 16-QAM technology, which has already been announced by a number of industry suppliers. The natural evolution of this idea would take the industry to 64-QAM and beyond – although there is a trade-off between modulation complexity and reach that could lead to some novel diversions.
But is it really this easy to predict the path the industry will take? In a word, no. Conquering metro networks will not be this simple. New technology paradigms usually arise in the low-cost metro first and then spread into core networks. A move of coherent core technology into the metro segment would be an exception to the rule, rather than a logical consequence.
First, sophisticated coherent technology based on flexible-rate interfaces simply cannot compete with classical 10G optics in terms of costs. Since most metro networks are relatively simple and fibre-rich, introducing a proprietary (read: expensive) technology is a direction only selected Tier 1 carriers are likely to take. It would also represent a complete about-face from the XFPs and SFP+ pluggable optical modules of today.
Second, the higher price per bit of coherent technology is inevitably tied to its greater power dissipation and physical footprint. Optical coherent interfaces are going through a similar development curve as did 10G, with a move from discrete components to pluggable modules, such as the CFP2-ACO. Coherent digital signal processing (DSP) chips – the brains behind the complicated unravelling of information at the receiver – are also steadily becoming smaller and more efficient.
But is it enough? Will the resulting coherent module power, footprint and prices be competitive against 10G inside the next year or so? The answer again is no – it is too soon. Instead of the huge 100G coherent metro surge in the next two years, it is more likely that we will see a controlled market introduction of coherent technology. Good enough to make money? Yes. But living up to the promise of transforming an industry? A far cry from it.
So where is the metro market headed? One trend that has been impossible to ignore is the emerging impact of digital media companies like Google and Facebook, with their ever-growing data centre capacity requirements and surging demand for internal and external connectivity. These companies not only tailor systems according to their needs, dissecting classical telecom offerings, but also begin to apply their own rules to the equipment sector, leading to shorter deployment cycles governed even more fiercely by cost, density and power consumption.
Products designed for the particular requirements of the data centre are emerging, but their mere existence highlights a basic uncertainty about the needs of the market. Platforms to please the data centre mind-set include a stripped-down, pizza-box size element performing one single function – the multiplexing of Ethernet onto an optical fibre. Other examples include optics multi-source agreements (MSAs) such as CWDM4, PSM4, CLR4 or OpenOptics, which promise to deliver much lower cost 500m- to 2km-reach transceivers compared to the classical 10-km LR4 interface.
Here, in particular, silicon photonics start-ups are attempting to challenge established players. And this is where datacom technology starts to spill over into the telecom sector. For example the OpenOptics alliance is developing its 2km-reach 4 x 28G silicon photonics-based optics at 1550nm, and could apply this technology to the line-side for shorter, sub-80-km, single-span interconnects. A robust direct-detection approach to higher speeds would be possible, for example, when combined with 4-level pulse amplitude modulation (PAM4). PAM4 devices are being introduced to cover applications as wide as active optical cables, backplane connectivity, or consumer cables, and could spill over onto coloured optical interfaces.
So what is a possible future scenario? Coherent flexible-rate is one option in this evolving landscape. However, we believe a combination of direct-detect and coherent interfaces is required to address the whole of the market, and the future dynamics of these technologies cannot be easily predicted. In metro applications, silicon photonics would enable highly-parallelised, inexpensive direct-detect interfaces with little or no DSP processing required.
If direct-detect technology can get ahead of the coherent curve, we might see a shift in network architectures. For the majority of applications, metro networks could be simplified with point-to-point, single-span connectivity with electrical switching and grooming at the junction nodes, without requiring any sophisticated optical infrastructure or its accompanying network management. Anyone who was hoping to see 200G 16-QAM and Raman amplification in the metro will probably be disappointed.
Ultimately, the data centre mind-set could, we believe, spill over into and completely change a significant portion of classical metro network architectures. Thus, a direct-detect pluggable may well be the future of metro 100G and beyond.
- Dr. Maxim Kuschnerov is a product manager for DWDM technology at Coriant, Munich, responsible for high-speed interfaces and photonic layer technology