Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3_B400G] 10km over duplex SMF objective



I support splitting the objectives.

Up to 10km LR1 PMD could both be used for intra-datacenter applications.  Low latency implementation is very important for intra-DC applications.  The choice of LR4 or LR1 depends not only on the reach, but also on the system architecture.   For example, operators choose FR4 or DR4 implementations not entirely just based on the link budget and reach, but also based on the fanout and switching requirements in the overall system design.
 
--
Cedric F. Lam
Cell: +1 (949) 351-2766


On Wed, Feb 15, 2023 at 2:31 PM Eric Maniloff <eric.maniloff.ieee@xxxxxxxxx> wrote:

Hi Chris,


I’m supportive of this modification to the 802.3dj objectives. I reviewed the CSD to make sure the applications we’re addressing by these are included, and this seems to be covered already.

 

There have been some comments that 800G LR4 and LR1 will be addressing different applications. I disagree with this. There may be some applications only addressed by one of these, but there will potentially be significant overlap. Both PHYs are likely to have applications inside the data centers and for campus applications – they will have similar pressure on latency and power since they are being applied on the same media and reach. 

I had a discussion with John D’Ambrosia on the topic of distinct identity. He pointed me to the following reference:

 

https://www.ieee802.org/3/cz/public/30_nov_2021/dambrosia_3cz_01_301121.pdf#page=7

 

Defining these on different numbers of wavelengths meets distinct identity.

 

So I’m in favor of separating these objectives, and letting us develop optimal specs for both coherent and IMDD objectives.


Regards,  Eric


On Mon, Feb 13, 2023 at 7:40 PM Chris Cole <chris@xxxxxxxxxxxxxxx> wrote:
Dear 802.3dj Colleagues,

During last week's meeting, there were a number of excellent presentations which gave us a clearer picture of the 10km over duplex SMF objective. 

Williams showed that the IMDD LR4 solution can result in the lowest cost if it leverages high volume DR4 and FR4. (Spoiler alert:  DR4 and FR4 PHYs will support LR4.) He further showed that the Coherent LR1 solution can have an unallocated link budget which can be used to support greater reach. We know that LR1 can easily have an even greater unallocated link budget which can be used for high-loss short-reach intra-datacenter applications like optical switching. This suggests that in the Task Force there are multiple important constituencies which will not be well served by one solution. 

Other presentations showed that there is important technical work that needs to be done and decisions that need to be made. An IMDD example is the FWM penalty. A Coherent example is O-band vs. C-band. The IMDD and Coherent technical issues and decisions are distinct, and their investigations are decoupled. 
After discussing this and process issues with our distinguished Chair, John D’Ambrosia, we would like to propose to the Task Force that we replace the following objective:

Define a physical layer specification that supports 800 Gb/s operation:

·       over a single SMF in each direction with lengths up to at least 10 km,

with the following:

Define a physical layer specification that supports 800 Gb/s operation:

·       over 1 wavelength over a single SMF in each direction with lengths up to at least 10 km,

·       over 4 wavelengths over a single SMF in each direction with lengths up to at least 10 km.

These two objectives are distinctly different, and examples can be found in prior projects, as well in 802.3dj itself, which has objectives targeting 800 GbE 2km operation over either 4 parallel fibers or 4 wavelengths.

We would like to get your feedback on this approach, and incorporate received comments into a proposal to be made during the March Plenary meeting.

Thank you

Chris




To unsubscribe from the STDS-802-3-B400G list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1


To unsubscribe from the STDS-802-3-B400G list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1


To unsubscribe from the STDS-802-3-B400G list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1