Thanks for your comments.
I always value your opinion and you may be right; in fact I completely agree with your comment “We are much better at writing specs than we are at predicting markets.”
However, I don’t recall ever pruning a data rate in a SG – we prune options and we make choices but
if 200G is in the CFI scope, it is highly likely that it will be in the objectives. It will not be pruned in the SG. We may only select one 400G option but there is
only one 200G option on duplex fiber: SR1.4. It still seems to me like we are trying to legitimize a solution through IEEE.
Let’s take a recent example, 802.3cd:
“50 Gb/s Ethernet Over a
Single Lane and Next Generation 100 Gb/s and 200 Gb/s Ethernet Study Group”
Sure, we pruned options but there is an objective for 50G, 100G and 200G.
We have two MSAs for 4 wavelength SWDM, one for 40G and one for 100G. How do we decide what to take to MSA
and what to take to IEEE? While we did address 100GBASE-SR1.2 in 802.3cd and rejected it, we never addressed 100GBASE-SR1.4 in 802.3cd. Why? More importantly, why not add PAM4 to the 100G SWDM4 MSA for 200G and call it a day?
I think the decision to include 200G will bog down the SG but that may just be one of the issues. I am primarily
interested in 400G over fewer pairs and specifically 400GBASE-SR4.2 but I am just one person in the committee. I made that decision based on what has been standardized to date – we have a solution already standardized at 200G and I am not a big fan of the
multiple options addressing the same application space. We have yet to define a duplex multimode solution in IEEE after 10G – I am not sure why we should start at 200G. It makes no sense to me.
Respectively,
Steve
Steven E. Swanson
Senior Standards Manager
Distinguished Associate
Global Technology & Industry Standards
MT&E
Corning Optical Communications
800 17th Street NW
Hickory, NC 28603-0489
t 828-901-5328
t 607-974-5757
m 607-725-1129
swansonse@xxxxxxxxxxx
Standards are a bridge between markets and technologies; whoever controls the bridge controls the future…
From: Vipul Bhatt [mailto:vipul.bhatt@xxxxxxxxxxx]
Sent: Wednesday, October 18, 2017 7:51 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Announcing NEA Ad Hoc call on Next-Gen MMF PMDs from noon to 1pm US Eastern time on 10/13 and 10/27
Steve,
The “dead end” is not a good argument. Even if it were true, it doesn’t matter to this CFI, and in fact it is not true. Please see attached table. We see a long and prosperous future ahead for both duplex and parallel MMF solutions.
To the larger point: It is unwise to throw away the 200G option from the scope prematurely. There is enough uncertainty about it to warrant inclusion, not exclusion.
Whether we consider a 2:1 or a 4:1 upgrade ratio in Server-Tier1-Tier2 speed hierarchy for all market segments over time (not just large data centers), it gets harder to be sure 200G doesn’t belong anywhere in those combinations. There
will be 6.4T switch versions with 50G lane I/O, making it possible for 32x200G QSFP56 solutions to be available in a 1U box. Or 12.8T switches with 64x200G QSFP56 in a 2U box. Market direction may well depend on the time gap between the maturity of 100G optics
(happening now) and the maturity of 400G optics.
I have some sympathy for the “we don’t see a demand today” argument. But let’s be careful. We are much better at writing specs than we are at predicting markets. There is definitely doubt here, for and against. The question is how best
to make a judgment call.
When in doubt, the CFI scope should be more inclusive and less exclusive. The pruning should be left to the objectives-selection process because it’s a much better process for it – meetings, presentations, straw polls, motions, and so on.
We need such a robust and pluralistic process to make a judgment call on what to leave out.
Best Regards,
Vipul
Vipul Bhatt
408-461-8521 mobile
From: Dale Murray - LightCounting [mailto:dale@xxxxxxxxxxxxxxxxx]
Sent: Wednesday, October 18, 2017 12:38 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: [EXTERNAL]: Re: [802.3_NGECDC] Announcing NEA Ad Hoc call on Next-Gen MMF PMDs from noon to 1pm US Eastern time on 10/13 and 10/27
Steve,
I think we are going in circles, but I’ll respond below.
Dale
From: Swanson, Steven E [mailto:SwansonSE@xxxxxxxxxxx]
Sent: Wednesday, October 18, 2017 9:33 AM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Announcing NEA Ad Hoc call on Next-Gen MMF PMDs from noon to 1pm US Eastern time on 10/13 and 10/27
Robert,
I have given the CFI some more thought and I am still struggling with the inclusion of 200G in the scope of the NGMMF CFI. I have developed a short deck (attached)
that highlights my concerns. As you know, I am a huge supporter of MMF solutions and believe that MMF has lots to offer the IEEE application community going forward but remain concerned with the continued “promotion” of a duplex plant for short reach applications.
•
The case for including 200G over fewer MMF pairs in the CFI is not clear That’s why we have a
study group phase.
–
Is it an upgrade path for proprietary solutions on existing duplex cabling? It certainly could
be. Why would that be a problem? I previously gave the example of 400G‑DR4 being an upgrade for proprietary 100G-PSM4. Don’t people want to transition to standardized solutions when they have a choice?
–
Do customers want to deploy 200G straight away?
If “straight away” means right now, then of course the answer is no. No one is trying to deploy any 200G solution right now.
–
I see no upgrade path past 200G on duplex cabling
OK, but we’ve never brought such an argument against any other PMD. How is this a negative against 200G on MMF duplex?
–
The identified solutions cannot support breakout This is correct, but 200G optical will be mostly
switch-to-switch so why is breakout a requirement?
•
It is for these reasons that I cannot support 200G in the scope of the NGMMF CFI
–
I don’t think anything will come to light in the SG that will change this view
–
But I remain open to counter arguments
At this point, I will not be able to support the CFI as currently worded. I could fully support the CFI if it was focused on 400G over fewer MMF pairs which is really
what I think we are really trying to address anyway (an alternative to 400GBASE-SR16).
Best regards,
Steve
Steven E. Swanson
Senior Standards Manager
Distinguished Associate
Global Technology & Industry Standards
MT&E
Corning Optical Communications
800 17th Street NW
Hickory, NC 28603-0489
t 828-901-5328
t 607-974-5757
m 607-725-1129
Standards are a bridge between markets and technologies; whoever controls the bridge controls the future…
From: Lingle, Robert L (Robert) [mailto:rlingle@xxxxxxxxxxxxx]
Sent: Wednesday, October 11, 2017 8:14 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Announcing NEA Ad Hoc call on Next-Gen MMF PMDs from noon to 1pm US Eastern time on 10/13 and 10/27
Dear Dale, thanks for those useful comments.
Dear Steve, thanks for expressing your concerns in detail. I am delighted that you fully support the 400G aspect of the CFI! I encourage you to bring your concerns over 200G duplex to the Study Group that I hope will meet in January, for
a full and fair hearing.
If I followed your suggestion, then you and I would effectively be making the decision for everyone about whether 802.3 will consider 200G duplex PMDs.
If I keep 200G duplex in scope – as suggested by the Wells Fargo use-case shown in the CFI draft from Charlotte (slide 16 in link) – then I am allowing a 75% vote in the Study Group to determine what 802.3 should do.
http://www.ieee802.org/3/ad_hoc/ngrates/public/17_09/lingle_nea_01_0917.pdf
To me the choice between those two options is clear: we should form a Study Group and hear presentations that are both for and against 200G duplex, then vote in March!
The gentleman from Wells Fargo who attended the NEA ad hoc session in Charlotte has expressed interest in contributing additional material regarding the application of 200G duplex over MMF during the Study Group. I think many people would
want to see that. You may wish to make a contribution against an objective for 200G duplex PMDs. Should we deny the community the opportunity to hear this debate?
I hope you can be persuaded that this is the right approach within our process, with a greater chance of leading to the optimal outcome for the Ethernet ecosystem. In the meantime, I have removed your name as a supporter, per your note
below.
As to your detailed points, Dale Murray has expressed well many of my specific thoughts on why we should consider 200G duplex in the Study Group.
With regard to OM5, I have reduced the references to two (2) in the draft we will review Friday (linked below); this is the minimum. MMF progressed from OM1 to OM2 to OM3 to OM4 as we moved from LEDs to VCSELs and higher speeds. This
is the progression of technology. In that same vein, it runs counter to all indications to pretend that WDM and OM5 have no role to play within the scope of this CFI. TIA completed standardization of OM5 by 2016 with no dissenting votes. All major MMF fiber
and structured cabling companies voted “yes.” OM5 MMF has also successfully progressed through the international fiber & structured-cabling standards bodies, including IEC 60793-2-10ed6 where OM5 was approved with no dissenting votes. OM5 has a natural role
to play as multiple wavelengths are used over MMF. Ethernet is not a platform for promoting any technology for its own sake, but it is a place where technology evolves and is codified. It is our responsibility to recognize and consider advancements that are
relevant to this CFI.
http://www.ieee802.org/3/ad_hoc/ngrates/public/calls/17_1013/lingle_nea_01_171013.pdf
Warm regards,
Robert
Robert Lingle, Jr., Ph.D.
Director, Market & Technology Strategy
Your Optical Fiber Solutions Partner™
2000 Northeast Expy | Norcross, GA 30071
Office: 770-798-5015
Mobile: 404-886-3581
From: Dale Murray - LightCounting [mailto:dale@xxxxxxxxxxxxxxxxx]
Sent: Tuesday, October 10, 2017 11:54 AM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Announcing NEA Ad Hoc call on Next-Gen MMF PMDs from noon to 1pm US Eastern time on 10/13 and 10/27
Steve,
I also questioned whether the CFI effort should include a 200G duplex MMF solution and a while back I spoke with Robert about it. He had at least one request for it to be included and he, as
the leader of this effort, has chosen to let the process play out rather than screen out objectives prior to becoming a study group.
I have since spoken with a vendor who sees a need for a 200G MMF duplex solution. While I am not yet clear on the market opportunity it might have, I guess it depends on whether 200GbE ever
becomes an enterprise speed. I do see the lag shrinking between cloud and enterprise so it could happen. I also note the MMF duplex solutions are mostly for brownfield applications so they need to be considered independently of SR4 solutions. Having both is
not a conflict.
You correctly note that the IEEE did not support a duplex solution at 40G and 100G, but the need was there and the need is being satisfied by both proprietary and MSA efforts. As to not having
40G and 100G MMF duplex solutions, I note that we also did not standardize PSM4 at 40G or at 100G, but we are now doing so at 400G.
The decisions made in 802.3ba were done in 2008 and much has changed since then.
You correctly note that 802.3cd rejected a duplex 100G MMF solution (SR1.2) and instead chose an SR2 solution. It was my impression that .3cd prioritized breakout opportunity over brownfield
MMF duplex. Will we see the applications that were suggested for SR2? I’m not sure. We will see deployment of both proprietary and MSA-based 100G duplex MMF solutions and we at LightCounting project a good future for this product category. So I respectfully
disagree that because duplex solutions were not standardized at 40G and 100G that sets an unchangeable precedent.
As to OM5 reflected in the CFI deck, it’s has been reduced considerably to mostly “FYI” material. Reach is important, especially in multimode so it’s not wrong to suggest there is a new technology
that could extend reach. The poll on reach you cite from 802.3ba was many years ago when data centers were much smaller. Since then, extended short reach (eSR) modules have been popular at 40G and this will repeat at 25G and 100G.
At this stage of the process, we need to get a study group formed to consider next-gen MMF objectives. If an objective is proposed in a study group for 200G MMF duplex, it will need to be defended
and achieve the necessary vote threshold. The process will play itself out.
Dale Murray
From: Swanson, Steven E [mailto:SwansonSE@xxxxxxxxxxx]
Sent: Tuesday, October 10, 2017 12:11 AM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Announcing NEA Ad Hoc call on Next-Gen MMF PMDs from noon to 1pm US Eastern time on 10/13 and 10/27
Robert,
In advance of the CFI meeting in Orlando, I would like to comment on the CFI wording. I fully support the work on NG MMF but cannot support the proposed wording of
the motion. I am asking the group to consider modifying the wording to focus on 400 Gb/s PHYs over Fewer MMF Pairs than Existing Ethernet Projects and removing 200 Gb/s.
My rationale is as follows:
There is motivation to support a 400G solution over fewer pairs as an alternative to 400GBASE-SR16.
We are standardizing 200GBASE-SR4 in 802.3cd and there is no precedent for supporting a duplex solution at 200G – at least it was not supported at 40G or 100G – so
why are we now thinking that it will be supported at 200G?
In IEEE 802.3ba, we defined 40GBASE-SR4; a proposal for a duplex solution was rejected by 802.3ba
·
It was noted that a 4 lane parallel short wavelength based PMD is the highest density, lowest power consumption and lowest cost 100m solution and I don’t think that has changed.
·
A straw poll to consider a duplex option yielded the result to develop a duplex based standard outside of 802.3ba by more than a 2 to 1 margin; several voted for the case that no
duplex MMF solution was needed.
·
40GBASE-SR4 was adopted 85-0
In IEEE 802.3bm, we defined 100GBASE-SR4; a proposal for a duplex solution was not proposed in 802.3bm
·
100GBASE-SR4 approved
·
To my recollection, no 100G MMF duplex option was proposed
In IEEE 802.3cd, we are defining 100GBASE-SR2 and 200GBASE-SR4; a proposal for a duplex solution (100GBASE-SR1.2) was rejected by 802.3cd
·
Both a parallel and a duplex solution were considered in the 100G work
·
A straw poll showed support for the parallel solution over the duplex solution by a 3 to 1 margin
·
A subsequent motion to standardized the parallel solution passed by a 6 to 1 margin
There seems to be no reason to now support a duplex 200G solution when we have not standardized a duplex solution at 40G duplex nor a duplex solution at 100G.
Regarding the inclusion of OM5 in the CFI, I think it is a distraction that takes away from the main point of the CFI. It is clear that
all perceived solutions can be supported on OM3 and OM4 and while some are purporting support for a 150m link length on OM5, this does not seem to be a requirement based on a poll in 802.3ba:
The goal for extending the reach on MMF should be:
A. at least 150m on OM3
B. at least 250m on OM4
C. both A and B
D. at least 200m on OM4
E. both A and D
F. there should be no goal for extended reach
Result:
A: 7
B: 3
C: 5
D: 1
E: 0
F: 27
It is for this reason that I would remove the slides on OM5; the inclusion of a 200G solution seems like an attempt to legitimize a certain transceiver/fiber combination
(4 wavelength SWDM over OM5) when it is not really needed.
So, to summarize, I will fully support the CFI if it is worded as follows:
Motion at Closing Plenary:
Move that the IEEE 802.3 Ethernet Working Group authorizes the formation of a study group to develop a Project Authorization Request (PAR) and Criteria for Standards
Development (CSD) responses for "Next-generation 400 Gb/s PHYs over Fewer MMF Pairs than Existing Ethernet Projects"
I will withdraw my support of the CFI if it is worded as follows:
Motion at Closing Plenary:
Move that the IEEE 802.3 Ethernet Working Group authorizes the formation of a study group to develop a Project Authorization Request (PAR) and Criteria for Standards
Development (CSD) responses for "Next-generation 200 Gb/s and 400 Gb/s PHYs over Fewer MMF Pairs than Existing Ethernet Projects"
Best regards,
Steve
Steven E. Swanson
Senior Standards Manager
Distinguished Associate
Global Technology & Industry Standards
MT&E
Corning Optical Communications
800 17th Street NW
Hickory, NC 28603-0489
t 828-901-5328
t 607-974-5757
m 607-725-1129
Standards are a bridge between markets and technologies; whoever controls the bridge controls the future…
From: Lingle, Robert L (Robert) [mailto:rlingle@xxxxxxxxxxxxx]
Sent: Friday, September 22, 2017 4:52 PM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: [802.3_NGECDC] Announcing NEA Ad Hoc call on Next-Gen MMF PMDs from noon to 1pm US Eastern time on 10/13 and 10/27
All,
We will hold two telecons related to Next-Gen MMF PMDs from Noon to 1pm Eastern US time on Friday October 13th and Friday October 27th.
The agenda for the first call will be to discuss the wording of a motion to authorize a Study Group to be made at the closing Plenary in Orlando.
We will also discuss any final tweaks to the CFI deck.
The second call will only be held if there is any business not finished on the first call.
Call-in and web information below
|
Level (3) |
Click below and we will call you to join the meeting:
- Or -
1. Dial-In:
- BRAZIL: 08008916743
- CANADA: 8007682983
- CHINA UNIFIED: 8008700602
- DENMARK: 80703123
- GERMANY: 08000004390
- INDIA: 18002090185
- IRELAND: 1800944107
- ISRAEL: 1809246050
- JAPAN: 00531190033
- UNITED KINGDOM: 08004960980
- UNITED STATES: 8007682983
- UNITED STATES (Toll): 2122313884
2. Enter Access Code: 7985015
3. Web Login:
https://core.callinfo.com/prt?an=8007682983&ac=7985015
Click here for a list of global toll-free numbers
Robert Lingle, Jr., Ph.D.
Director, Market & Technology Strategy
Your Optical Fiber Solutions Partner™
2000 Northeast Expy | Norcross, GA 30071
Office: 770-798-5015
Mobile: 404-886-3581