Are Two Carriers Better than One?

The world of optical networking transmission solutions divides into two main camps: power-density-cost-optimized and performance-optimized.

Power-density-cost-optimized solutions use standard pluggable technologies. Featuring strong (just not extraordinary) performance, they are the industry workhorses for most applications. These solutions will shine later this year with the commercial availability of 400GZR/ZR+ pluggables.

Performance-optimized solutions are the industry thoroughbreds. They use proprietary technologies and have the goal of maximizing bits throughput for a given fiber condition. They are used in applications that need to transmit that “extra mile”, such as DCI (data center interconnect) and undersea. This blog deals with this class of solution.

The most recent achievement heralded by performance-optimized solutions is commercial availability of “800G in a single wavelength”. This made me stop and think, why didn’t the announcement say, “800G in a single channel”. Similar, but not quite the same thing. Then it became clear. There are two competing approaches to achieving 800G in the industry today, we’ll call A and B.

Both approaches use a single optical channel. The differences have to do with optical carriers and optimum channel widths. Approach A uses two powerful carriers, each capable of achieving up to 600G, and a flexible channel width up to 150GHz. Approach B uses a single super-powerful carrier, capable of achieving up to 800G, and an invariable channel width of just over 100GHz. An analogy in may be found in speedboat design. Do you put all your power into a single engine (approach B) or split it into two engines (approach A).

Both approaches can do 800G transmission in a single channel. To compare them, we look at two practical criteria:

1. Their ability to transport the newest generation of high-speed 400GbE clients
2. how well they do so for brownfield fixed grid and greenfield flexible grid network applications.

We’ll start with brownfield. This application is important, because many if not most service providers want to continue employing their existing optical line systems that use 50GHz or 100GHz fixed spacing grids. Approach A is able to handle these spacings because it uses two carriers, each of which can fit into a fixed 50GHz channel. Two 400G carriers can be used to transport two 400GbE clients for about 200 km, or two 200G carriers to transport a single 400GbE client for over 1000 km. Approach B, however, cannot support brownfield deployments at all, since it is based on a single carrier with an invariable channel width greater than 100GHz.

Turning to greenfield, here each approach has different merits depending on the channel width used. At approach A’s preferred channel width of 150GHz, approach A features extraordinary performance, with an ability to transport two 400GbE clients about 2.5 times further in distance than approach B. Technically, this is because the OSNR for two 400G carriers within 150GHz is far less than the OSNR of a single 800G carrier at just over 100GHz.

Alternatively, approach B performs marginally better (about 20% further in distance) at its optimum channel width of just over 100GHz. This also means, because this is a narrower channel width, that approach B is spectrally more efficient.

There is an important lesson here as we approach the Shannon limit, where each increment in optical rates is more difficult to achieve. The lesson is that we need to go beyond the headlines and dig into practical use cases to understand the relative advantages and disadvantages of different approaches.

Table: Summary of two approach’s abilities to transport 2x400GbE clients or 8x100GbE clients at 800G in a single optical channel