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

Re: [802.3_100G-OPTX] P802.3cu Interim minutes posted

Hi Paul

 

The perspective was stated by multiple attendees, however it takes more than one meeting to reach new consensus. That is why the suggestion was made to hold off voting on the designation. This is primarily an end user issue, so participants need time to talk to their customers. Hopefully this will happen by the July meeting.

 

We discussed multi-lane, multi-wavelength PMD designation in 802.3bs in the context of 400G PSM4, and used it first in 802.3cm for 400G SR4.2. Going forward, we need 802.3 wide agreement on when the full designation can be shortened. My suggestion is that as long as there is no conflict between specific PMDs, we use the shortest possible designation. When there is a conflict, we use a designation that distinguishes between the PMDs and the generic designation, as done in 802.3ck.

 

Chris

 

From: Paul Kolesar <pkolesar098@xxxxxxxxx>
Sent: Friday, June 07, 2019 7:41 AM
To: STDS-802-3-100G-OPTX@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100G-OPTX] P802.3cu Interim minutes posted

 

Chris,

thanks for bringing this up.  I agree with you, and wish you could have verbally stated your perspective at the meeting in Salt Lake City. 

Regards,

Paul  

 

On Fri, Jun 7, 2019 at 12:56 AM Chris Cole <chris.cole@xxxxxxxxxxx> wrote:

Dear 802.3cu Task Force Participants,

 

During the May 2019 interim meeting, 100GBASE-FR and 100GBASE-LR were adopted as names for 100G single optical lane PMDs. The alternative not chosen was 100GBASE-FR1 and 100GBASE-LR1.

 

This is exactly opposite of the 802.3ck Task Force, which at the January 2019 interim meeting adopted 100GBASE-CR1 and 100GBASE-KR1 as names for 100G single copper lane PMDs. The alternative not chosen was 100GBASE-CR and 100GBASE-KR. We now have inconsistent naming conventions in 802.3, which misses the point of having standards. 

 

During the May meeting, I suggested that .3cu participants refrain from making this decision, to allow time to understand end user preferences, most importantly those of their datacenter operators. This is not a technical spec, so what matters most is operational convenience.

 

For single lane optics, having no digit after the letters made sense when there was no multi-lane alternatives. 10GBASE-LR and -SR were perfectly clear because there was no LR4 and SR4. The 10G four lane standard used different encoding so it was named 10GBASE-LX4, which meant there was no confusion with LR.

 

However, now that we have both single and multi-lane solutions, not using a 1 is confusing. The DR, FR, LR designation is used generically for any PMD with that approximate reach. If it is also used for a single lane PMD there is an ambiguity whether the reference is generic or specific. The table below shows this for FR and LR naming alternatives. Alt. A, which was adopted in May, results in an ambiguity between referring to the generic PMD with any lane count, and specific PMD with a single lane count.

 

PMD lane

count

2km alt.  A

(adopted)

2km alt. B

10km alt.  A

(adopted)

10km alt. B

generic

(any lane count)

FR

FR

LR

LR

1

FR

FR1

LR

LR1

2

FR2

FR2

LR2

LR2

4

FR4

FR4

LR4

LR4

8

FR8

FR8

LR8

LR8

 

Because what matters is the convenience of the end user, I asked volume end users of 100G optics about their preference, as committed to at the meeting. The table below has their positions.

 

End User

Single lane 500m, 2km, 10km

naming convention preference

Alibaba

DR1, FR1, LR1

Amazon

DR1, FR1, LR1

Apple

DR1, FR1, LR1

Facebook

DR1, FR1, LR1

Google

DR1, FR1, LR1

Microsoft

DR1, FR1, LR1

Tencent

DR1, FR1, LR1

 

There is no end user support for omitting the lane count from the name.

 

I will be raising this issue at the July meeting. I would like to suggest that everyone directly ask their customers what they prefer, so that we can be debate actual end user preferences, rather than aesthetics. We should also reconsider having contradictory naming conventions within 802.3.

 

Thank you

 

Chris

 

 

From: Mark Nowell (mnowell) <00000b59be7040a9-dmarc-request@xxxxxxxx>
Sent: Monday, May 27, 2019 9:25 AM
To: STDS-802-3-100G-OPTX@xxxxxxxxxxxxxxxxx
Subject:  [802.3_100G-OPTX] P802.3cu Interim minutes posted

 

Dear Colleagues,

 

I hope you all had safe travel home from Salt Lake City.

 

The minutes from last week’s meeting are posted here:

http://www.ieee802.org/3/cu/public/May19/

Regards,  Mark  

To unsubscribe from the STDS-802-3-100G-OPTX list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-100G-OPTX&A=1

To unsubscribe from the STDS-802-3-100G-OPTX list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-100G-OPTX&A=1