Toward the Long Haul: 400G and Data Center Interconnect

Among other factors, the explosion of cloud computing and the evolution toward the internet of things (IoT) and 5G have created tremendous pressure on data centers to ramp up both network capacity and the capacity to respond to traffic pattern unpredictability.  

This phenomenon is presently driving the industry to produce new solutions for DCI, or data center interconnect, and already has produced a number of promising trends. These span the DCI gamut, from short-range connections in and around data centers to metro to long-haul. Much of the current focus is on near-term adoption of 400 gigabit-per-second throughput (400G) schemes, while other data rates and technologies continue to play supporting roles in an evolving landscape.  

Within the short-range space, for example, the technology of pulse amplitude modulation (PAM) has risen to the forefront. PAM is an efficient and cost-effective alternative to the architecture-based approaches traditionally used to build greater capacity. An example is PAM4, through which optical transceivers can utilize one laser instead of four — and realize 100 gigabits per second throughput over a single fiber. The technology can be applied to existing architectures, serving to double bit rate. When amplification and dispersion compensation are incorporated into the system, PAM4 also can be used for longer distances.  

The long-haul space has also seen high-speed networking solutions being deployed proactively as a way of “future-proofing.” Although higher bit rates typically mean shorter ranges, overall connectivity is still increased by high-speed deployment. For instance, a 200G-capable DCI may only realize full speed within the smaller proximity of the metro — but can also operate at 100G for long-haul applications requiring increased optical reach. The same holds true for 400G deployments, where supporting 200G long-haul DCI network builds is a frequent reason for early 400G adoption.  

Another emergent capability seen throughout the data center ecosystem is programmability. Interfaces that can be adjusted to optimize capacity in response to real-time bandwidth demand increase operators’ ability to respond to unpredictability. Building in this kind of flexibility is also another way for operators to “future-proof” their networks. 

Although 400G deployment is still in its early stages, industry motivation to push it forward is strong. In addition to increasing traffic-carrying capacity and reducing operational costs, 400G offers a smaller footprint and lower energy consumption when compared to older data-rate schemes. It’s a technology developed as a response to tremendous pressure, which offers tremendous promise in return.