Work to define a 400 Gbit/s Ethernet standard formally begins in March, says John D'Ambrosia, outlining some of the key issues ahead.
It has been one year since the initial call for interest to start work on a 400 Gbit/s Ethernet standard. At the IEEE 802 Plenary in Beijing the week of March 16, the IEEE 802.3 400 Gbit/s Ethernet Study Group will be requesting approval of its project documentation, so the real work on the 400GbE standard can begin.
A review of the project documentation will provide insight into the physical layer specifications that will be worked on for the initial version of the 400GbE standard. There will be other 400GbE projects eventually. The key part is that this initial effort will spend significant time and energy on defining an architecture that will be flexible and enable future 400GbE implementations.
From a top view, 400GbE for full-duplex operation only will be defined where the Ethernet frame format using the beloved Ethernet media access controller will be preserved. It also will support the currently defined minimum and maximum frame size of the existing IEEE 802.3 Ethernet standard. Finally, support for optical transport network and energy-efficient Ethernet will be defined.
The proposed standard aims to define physical layer specifications that support link distances of:
- At least 100 m over multimode fiber
- At least 500 m over singlemode fiber
- At least 2 km over singlemode fiber
- At least 10 km over singlemode fiber
However, as few as two physical-layer specifications may be created, one targeting multimode fiber and another targeting the other three singlemode fiber objectives. The 500m objective may use a parallel fiber solution that could enable the breakout cable function that is popular today where a 40GbE cable is used to support four ports of 10GbE. In addition, a 2km specific solution might be less expensive than a 10km solution.
Diving into the technical details of each of these anticipated proposals will be part of the fun that awaits those ready for this project. It should be pointed out that physical layer specifications will be defined to support a BER of 10-13, which is an improvement over the 10-12 BER that was specified for 10GbE, 40GbE, and 100GbE.
Finally, the Study Group has agreed to define electrical chip-to-chip and chip-to-module application interfaces. I personally believe that the decision to develop these same interfaces for 40GbE and 100GbE played a key role in accelerating the adoption of 40GbE and 100GbE when tough debates occurred in the IEEE 802.3ba Task Force with the key stakeholders present. The developers of the QSFP and CFP families were probably happy that the IEEE 802.3ba Task Force tackled these specifications.
Industry discussions exploring the 400GbE architecture are already underway. Consensus-building forums, such as the Ethernet Alliance's 400GbE Subcommittee, will be crucial to the development of 400GbE. As chair of multiple IEEE 802.3 task forces, I have always told the individuals in these groups that the real work in developing a standard starts, not stops, when they leave the meeting room.
Stay tuned. Ethernet will be getting very interesting in the near future as the 400GbE architecture is debated.
— John D'Ambrosia is acting chair of the 802.3 IEEE 400 Gigabit Ethernet Study Group, chairman of the Ethernet Alliance, and an Ethernet evangelist and distinguished engineer at Dell.