400G everywhere: A universal currency for IP-optical convergence
We often speak about exploding traffic growth in terms of annual percentages. Driven by the world's insatiable appetite for on-demand content and cloud applications, traffic demand grows continuously, exponentially and, as we’ve learned in 2020, sometimes unexpectedly. But that's not quite how the networks that carry this traffic tend to evolve. The underlying “speeds and feeds” at which network equipment connects and operates grow predictably, even slowly — and at a glacial pace compared to the software innovations that run on top of them. The data-carrying capacity, or speed, of router and switch ports is gated by the development, standardization and commercialization of Ethernet — Gigabit Ethernet in 1998, 10 Gigabit Ethernet in 2002, 100 GigE in 2010 and now 400GE.
Each new rate becomes the currency of high-speed network connectivity and services for a generation, and its adoption the catalyst for a new network investment cycle. Fed by silicon advances that have enabled massive increases in switching capacity, routers are bursting at the seams, limited only by their ability to move data into and out of their switch fabric and connect to one another at high speeds over distance. By combining a fourfold increase in port bandwidth with advances in coherent electro-optics, 400GE promises to unleash a new wave of network innovation that will transform the way IP-optical networks are built and operated.
IP and optical: Together at last?
The data and transport worlds have always had a contentious relationship. Back in the days when TDM ruled transport, IP bent to the will of optical, contorting itself into transport-friendly formats through PDH (E1/T1) and Packet-over-SONET/SDH router interfaces. ATM made a run at smoothing things over with data-oriented packet features and TDM-inspired transport. But it failed to succeed in joining two worlds that were increasingly interdependent but hopelessly distant when it came to hardware, protocols, operations and even culture.
Ironically, it was Ethernet — a technology developed to shuffle data around buildings — that bridged the gap between the data and transport worlds, ultimately proving to be the lingua franca that enabled routers to be seamlessly interconnected across the globe through standardized interfaces over optical transport networks. It's therefore no surprise that a new era of IP-optical integration is poised to emerge alongside the wide adoption of a new Ethernet rate.
DCOs will drive IP-optical integration
But we've seen these transitions before. Multiple generations of Ethernet have come and gone, accompanied by matching optical wavelength rates, without a major shift toward integrated IP-optical network architectures. The reasons for this are many, but one significant obstacle to such integration has been the elevated space and power footprint of WDM electro-optics. This has prohibited the advent of router-based, pluggable WDM optical transceivers until many years after widespread adoption of each new rate by packet switching devices, necessitating the use of discrete and dedicated optical transponders for WDM transport. However, recent advances in electro-optics, particularly silicon photonics, have bridged this gap by allowing WDM optics to fit within the space and power envelope of high-speed router port form factors. For the first time, these advances coincide with the new Ethernet rate’s initial adoption.
Recognizing the promise of this technological intersection, the Optical Internetworking Forum's 400ZR specification has put IP and optical on a path toward convergence. This specification has the potential to integrate the two worlds in the purest of manners – IP, wrapped in Ethernet, launched straight from a router port over an optical wavelength through a pluggable digital coherent optic, or DCO.
A DCO's most important attribute is that it can be hosted directly in the router, which removes the space requirements, power consumption and expense of a separate optical transponder. By fitting within the dominant 400GE pluggable form factors, DCOs enable routers to enjoy the same port density when used for WDM transmission as they do for short reach connectivity. It's a perfect fit for the burgeoning application of metro data center interconnect, where 400ZR-compliant pluggable transceivers will benefit from data center economies of scale, as well as a robust and diverse supplier ecosystem.
The massive savings in cost, space and power will initially come at the expense of performance. Despite tremendous advances, today's state-of-the-art electro-optics still confront significant technical challenges on their journey to pluggability, primarily because of space and power constraints. Yielding to the compromises inherent in miniaturization, 400ZR and its more capable yet non-standardized cousin ZR+ are suitable for metro distances only. Does this mean the 400GE revolution will stop at the city limits?
400G everywhere (and anywhere)
Fortunately, optical technology continues to excel at what it does best: transporting data across the globe at the lowest possible cost per bit. The optical industry has been steadily preparing to transport 400GE as coherent digital signal processors have evolved to support faster baud rates and higher-order modulation formats. Although the technology has been available for several years, deployments of 400G wavelengths have been tempered by lack of demand and limited reach. 400G wavelengths supported by earlier coherent generations have been limited to metro and regional applications. With client traffic mainly coming from 100GE router interfaces, most operators have been content to deploy cost-effective and far-reaching 200G wavelengths (carrying 2 x 100G clients) to grow network capacity.
400ZR borrows from previous generations of coherent technology, steering its implementation in favor of power and size. In contrast, new-generation interfaces such as the Nokia PSE-V are designed to maximize the spectral efficiency and capacity–reach product of a WDM wavelength, with the goal of minimizing cost per bit over extended distances. These high-performance coherent interfaces use the most sophisticated digital signal processing algorithms and advanced optics to achieve robust, 400G-optimized wavelength performance across long-haul, ultra-long-haul and subsea networks.
The multiple variants of coherent technology complement one another and are optimized around the economic constraints that predominate within their target network applications. The design trade-offs made by DCOs reflect the short distances, constrained space and power, and plentiful fiber available in metro areas. High-performance optics maximize reach and spectral efficiency where fiber is scarce and operations costly. As they architect 400GE-based networks in the coming years, operators will achieve both lowest cost and highest performance only by drawing from a diverse and complete coherent product portfolio such as Nokia WaveFabric Elements.
The dawn of the 400G IP-optical era
The 400G era dawns with the alignment of multiple innovations and technologies. 400 Gigabit Ethernet will trigger a router investment cycle and spur demand for 400G wavelengths. Cost- and space-efficient pluggable coherent transceivers will finally enable the physical integration of the IP and optical worlds, while the latest generation of high-performance coherent technology powers 400G transmission over any distance. This technological and temporal convergence is unique in the history of IP and optical networks and promises to make 400G the new universal network currency.