The Next Ethernet

In today’s world the importance of bandwidth to everyday life has become obvious to all. Given its presence throughout the networks that enable today’s world, it is urgent that Ethernet support the networks of today and tomorrow.

The IEEE 802.3 Ethernet Working Group has learned that even when it is not developing the next speed of Ethernet, it must keep a watchful eye on industry bandwidth trends in anticipation of when work on the next speed of Ethernet must begin. The latest IEEE 802.3 Ethernet bandwidth assessment was published in April 2020.¹ Figure 1 illustrates the findings of the forecasted bandwidth growth for a number of key application spaces from 2017 to 2025, which showed that, relative to 2017, these applications would grow 2.3X to 55.4X by 2025.

Figure 1. IEEE 802.3 2020 Ethernet Bandwidth Assessment.

Faced with this latest assessment, the IEEE 802.3 Beyond 400 Gb/s Ethernet Study Group was formed this January and began working on defining the project that will establish the next rate of Ethernet.

There are two key questions this group will need to answer to define this project:

What is the next rate or rates of Ethernet?
What physical layer specifications will be targeted?

As noted in Figure 1, two rates being discussed as likely candidates are 800 Gb/s and 1.6 Tb/s. While 800 Gigabit Ethernet (GbE) is seen as the nearer-term solution to develop, its 2X growth over 400GbE doesn’t keep up with any of the observed data rates. On the other hand, 1.6 Terabit Ethernet (TbE) represents a 4X leap over 400GbE, but would be a wider interface on foreseeable technology, not necessarily meeting all of the density demands of networks.

The rate debate will go hand-in-hand with the determination of the target physical layer specifications. Today’s Ethernet specifications address reaches from the centimeters of chip-to-chip interconnect to the 80-km reaches being developed now for DWDM networks. Given the breadth of copper, multimode fiber, and single-mode fiber specifications, the study group will need to prioritize its efforts and identify those specifications that the industry will need in the near term. While there is often a push by some to get their favorite physical layer specification adopted as part of a new speed project, the reality is a new rate of Ethernet develops a family of specifications that takes years to develop. For example, the initial effort to develop 100GbE started in 2006, and since then, there have been seven follow-up projects that expanded the 100GbE family, with three of them still underway.

The latest 100GbE projects all include efforts to develop serial 100-Gb/s signaling specifications for copper, multimode fiber, and single-mode fiber projects. It is expected that that these various solutions will be considered as part of the “toolbox” for developing the next speed of Ethernet. Others are looking to make the leap to 200-Gb/s signaling, both electrically and optically, to reduce the width of the potential solutions. The development of 200-Gb/s signaling would line up with creating a x4 solution for 800GbE, which has traditionally been a sweet spot in terms of cost and density for the Ethernet space. This would also enable a x8 solution that could address 1.6TbE. Furthermore, the scope of the study group would allow it to consider using any signaling technologies developed to support a higher speed to address existing rates of Ethernet. So, the introduction of 200-Gb/s signaling for any physical layer specification would allow the group to also consider developing the same physical layer specifications for 200GbE and 400GbE.

As the study group contemplates these issues, others will be looking ahead to start considering how to develop the underlying architecture that will enable the next rate of Ethernet. It will need to be flexible and be able to scale with future technology. As noted, there is already a debate about leveraging 100-Gb/s or 200-Gb/s signaling. In the future 400-Gb/s signaling may be the next “big thing” and the next rate architecture will need to be flexible enough to support its use.

Conclusion

So here we go again. The next rate of Ethernet. While some may pause and say 400GbE will solve all our problems, all one needs to do is look back and see how fast the industry blew through 40GbE. Some may say that the times are different, and they would be right! It is so much more complex with so many opportunities for applications to drive high bandwidth.

Join us on February 17, 2021 for the Lightwave webinar “The Next Ethernet – Moving Beyond 400 Gb/s,” where we will discuss these issues and engage in an industry discussion about its needs for the next rate of Ethernet. (Find out more here.)

John D’Ambrosia is chair of the IEEE 802.3 Beyond 400 Gb/s Ethernet Study Group and distinguished engineer at Futurewei Technologies.