Re: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

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

Dear 802.3 Colleagues,

I have a shocking and totally unexpected development to report.

There is at least one person who actually looked at the numbers in the Ethernet volume table.

They pointed out a math error which is corrected in the updated table below.

Also, here is further clarification on the λ categories which I made in a private exchange.

λ#/fiber = 1: SR, SR2, SR4, SR8, SR10, LRM, DR, PSM4, FR, LR, ER, ZR

λ#/fiber > 2: Duplex MMF, SR4.2, CWDM4, 4WDM10, 4WDM20, FR4, LR4, ER4

Chris

		Vendor Survey, Q1 + Q2 2020 Ethernet Data, 9/17/20
λ Type		10G λ	25G λ	50G λ	100G λ
Fiber	#/Fiber	Volume	Volume	Volume	Volume
SMF	= 1	5,139,912	1,693,126	28,325	596,158
	> 2	1,871,800	11,065,264	814,384	16,004
	> 1	7,011,712	12,758,390	842,709	612,162
MMF	= 1	8,927,088	8,713,686	787,200	-
	> 2	1,224,000	1,040,000	-	-
	> 1	10,151,088	9,753,686	787,200	-

From: Chris Cole <chris.cole@xxxxxxxxxxx>
Sent: Wednesday, September 23, 2020 5:33 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

Hi Ali, Steve,

To clarify the categorization, the arrangement of the table data into Gray & WDM λs (or 1 & 2 or more λs) vs. λ rates is not in the original LC report. LC reports volume in the traditional way, by module type. The math is not terribly difficult to derive former from the latter.

Chris

From: Chris Cole
Sent: Wednesday, September 23, 2020 3:14 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: RE: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

Hi Steve

You bring arguably the most important perspective, which is the network one. I will leave it to those smarter than me to explain it to the Group.

The perspective I am brining is of the underlying technology; i.e. several decks below where you are standing. While you pilot our great optics ship towards new horizons, this is a perspective from deep in the holds, where us grimy stokers are shoveling the coal into the furnaces to keep us moving forward.

Since the names appear to be getting in the way of looking at the numbers, below is the same table with hopefully less controversial descriptors.

Chris

From: Trowbridge, Steve (Nokia - US) <steve.trowbridge@xxxxxxxxx>
Sent: Wednesday, September 23, 2020 2:53 PM
To: Chris Cole <chris.cole@xxxxxxxxxxx>; STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: RE: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

Hi Chris,

Of course I know that Mr. Anslow would always cringe at use of the word “gray” with respect to optics, but your characterization is somewhat different from how the term would be used thinking from a network perspective.

Client interfaces are normally though of as gray, and what distinguishes this is that there is only one logical signal per fiber. It may be a multi-lane O-band Ethernet client plug, and I don’t actually care how many lanes it is (e.g., 400GBASE-FR4) – I can’t really use the rest of the fiber for anything else. Logically, it is a single, indivisible, 4-wavelength signal.

The WDM interfaces would be thought of as network or “line-side” interfaces with multiple signals per fiber (usually C-band, maybe C+L).

Regards,

Steve

From: Chris Cole <chris.cole@xxxxxxxxxxx>
Sent: Wednesday, September 23, 2020 3:42 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: [802.3_NGECDC] Fwd: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

Hi Ali,

Yes, it is a fascinating report; dispels many deeply held beliefs. 😊

WDM is any optic that has two or more wavelengths per fiber, on any wavelength grid.

Gray is any optic that has only one wavelength per fiber.

Chris

From: Ali Ghiasi <ali@xxxxxxxxxxxxxxxxx>
Sent: Wednesday, September 23, 2020 2:37 PM
To: Chris Cole <chris.cole@xxxxxxxxxxx>
Cc: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Fwd: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

Hello Chris,

Interesting report!

What is the definition of Gray and WDM? I assume LC must be counting CWDM4, 4WDM10, FR4, and LANWDM as WDM.

Thanks,
Ali Ghiasi

Begin forwarded message:

From: Chris Cole <chris.cole@xxxxxxxxxxx>

Subject: Re: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

Date: September 23, 2020 at 1:34:18 PM PDT

To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx

Reply-To: Chris Cole <chris.cole@xxxxxxxxxxx>

Our friends at Light-Counting just published their latest optics vendor survey report. As always it’s a treasure trove of wonderful data, and it’s irresistible to play with.

A problem in the way we look at volumes is focus on counting boxes, rather than underlying technology, for example the graph on page 57 of the below referenced CFI deck. Another example is just counting modules, whose limitations were illustrated by the recent discussion of break-out on this reflector.

Our Logic colleagues have figured how to get a true picture of switches. In describing an ASIC, not only is the total bandwidth of the BGA package listed, but also the number and speed of the of SerDes and MAC cores. Since we get some of our best technical ideas from the Logic guys (ex. KR4 and KP4 FECs), why not learn from their approach here.

The underlying technology in optics are the wavelengths, so why not count those as well. In the below table, their volume is aggregated across all Ethernet codes for the first half of this year. There are multiple interesting observations that arise from this perspective.

Those that have access to the LC report will find even more interesting observations in wavelength cost data averaged across all codes.

Chris

From: John D'Ambrosia <jdambrosia@xxxxxxxxx>
Sent: Sunday, September 20, 2020 5:58 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: [802.3_NGECDC] Draft - Beyond 400 GbE CFI Consensus Deck

All,

I have uploaded the latest version of the draft CFI Consensus Deck. Please see https://www.ieee802.org/3/ad_hoc/ngrates/public/calls/20_0928/dambrosia_nea_01_200928.pdf.

This is not the version intended for presentation at the 9/28 meeting. I will update the latest version as further modifications are made. Please note I have also updated the supporters list. If you emailed me your support, please check the list and confirm I have added you correctly.

Regards,

John D’Ambrosia

Champion, Beyond 400 GbE CFI Consensus

Chair, IEEE 802.3 NEA Ad hoc

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

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

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