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

Re: [802.3_NGMMF] what form for our objectives? please discuss



Mike & Paul, thanks for bringing out these excellent points. I will number my responses 1-4 same as Mike, and I add the overall project schedule as issue 0)

Issue 0) With regard to Beth's assessment of the schedule, I have not reverse-engineered it, but I assume it to be correct. This reality makes me even more desirous of moving to Task Force out of the March meeting if possible. I frequently get input that we need to move quickly on the 400G objective over lower fiber count to avoid proliferation of proprietary solutions. Just yesterday I received input that 200G over MMF (for any cable type) needs to move quickly to meet anticipated market window. Some may quibble with one of these statements, but I get strong push from multiple sources to press ahead on 400G for sure. So I favor attempting to meet the March deadline and making needed choices now that are feasible and required, while judiciously (within our process) leaving some to the Task Force.

Issue 1) With regard to the # of objectives for a given speed --> I believe we could define two sets of specs for the 400Gb/s case if we gain consensus that there exists BMP/EF/TF for two different cable types with fewer than 16 pairs. Hypothetically and for example, if participants brought forward convincing evidence & arguments for 400G over both 4 pair MMF and 1 pair MMF, which are both common in the installed base, with different connector types and relative costs, then I believe our project could generate two 400Gb/s objectives. For a 200Gb objective over fewer than 4 pairs MMF, I could only envision one variant - either over 2 pair or 1 pair MMF - as being justifiable. 

Issue 2) choice of data rate --> I agree we must make the choice of whether we will work on 200G and/or 400G in SG

Issue 3) number of fiber pairs --> I believe we can write objectives as "fewer than 16" and "fewer than 4" fiber pairs for 400Gb/s and 200 Gb/s, respectively, as Paul suggests.

Issue 4) re-use of existing FEC/logic/PCS --> I think we can & will show (Mark Gustlin contribution) that we will be able to define 50 Gb/s PMDs for 200Gb/s and 400 Gb/s that re-use existing FEC/logic/PCS clauses. If someone makes a strong case in Geneva that 802.3 should define a low latency FEC for MMF, then this debate will occur. However, I have not been told for sure that anyone will propose this, and I am seeking clarification. (I have not heard that anyone would propose stronger FEC for this project).  The options are to A) choose to define a new low latency FEC, B) choose not to pursue it and thereby strengthen our case for BMP & TF by stating definitively that we will re-use existing clauses, or 3) write the objective such that we can write more than one set of specifications for the speed, leaving open the possibility of adding a new FEC TBD in the TF.  People may have strong opinions on this?

Clarification of question Paul asked: I asked several past chairs about the difference between "Define a 400Gb/s PHY for operation over ..." and "Provide physical layer specifications which support 400Gb/s operation over ..."  Both agreed that either language allows definition of a new PHY.  The second wording does not limit one to a PMD-only project.  An essential difference is that first wording does limit one to defining only one PHY, whereas second case allows more than one.

Warm regards,
Robert

On Tue, Jan 2, 2018 at 5:11 PM, Kolesar, Paul <PKOLESAR@xxxxxxxxxxxxx> wrote:

Mike,

The interleaving of meetings of the upper layer committees and boards is certainly a consideration.  While the slide you included does not provide specific dates, I get the thrust that this interleaving is not favorable for a September meeting if the project approval happens in July rather than in March.

 

I need to correct my issue 4 assessment statement. I should have said TF rather than SG in last sentence, as corrected here.

Issue 4: new FEC or not new FEC

The case can be made to reuse the existing FEC, as you have begun.  The case could also be made, for example, to eliminate FEC to address the low-latency market.  I’d prefer to let this be decided in the TF phase, but in doing so the CSDs would need to remain agnostic about the latency aspects of the applications.

 

Paul

From: Dudek, Mike [mailto:Mike.Dudek@xxxxxxxxxx]
Sent: Tuesday, January 02, 2018 4:59 PM
To: Kolesar, Paul <PKOLESAR@xxxxxxxxxxxxx>; STDS-802-3-NGMMF@LISTSERV.IEEE.ORG
Subject: RE: what form for our objectives? please discuss

 

Email Security Warning:

The following message was sent from an external e-mail address. Exercise caution when opening attachments, clicking links, or exchanging information.

Beth’s slides are saying that if the PAR and CSD are not approved until the July Plenary then the “upper committees” dates mean that the Task Force won’t be approved in time for the Sept Plenary.   (See below).  

 

 

 

From: Kolesar, Paul [mailto:PKOLESAR@xxxxxxxxxxxxx]
Sent: Tuesday, January 02, 2018 1:14 PM
To: Dudek, Mike <Mike.Dudek@xxxxxxxxxx>; STDS-802-3-NGMMF@LISTSERV.IEEE.ORG
Subject: RE: what form for our objectives? please discuss

 

Mike,

Your deep dive into the subtleties is useful. You caught more of them than I did.

 

Issue 1: number of solutions per data rate.

The first two bulleted objective examples state a singular PHY.

The second two bullets could be interpreted as singular or multiple because the word “specifications” is plural.  Does that plurality mean multiple solutions or that even one solution has multiple specs?  If one solution per data rate were intended, this could be clarified by saying “Provide a physical layer specification …”

 

Issue 2: which data rate(s).

This choice could affect the CSDs, depending on how they were written.  I agree that this choice should be made by the SG.

 

Issue 3: number of fiber pairs.

This choice could affect the CSDs, depending on how the CSDs were written.  For example, the CSDs could be written such that the merits of fewer pairs could be stated generically without the need to know the number, or they could be written to describe applications that specific fewer-pair solutions would address.  I’d prefer the former approach, so the pair counts get decided in TF phase.

 

Issue 4: new FEC or not new FEC

The case can be made to reuse the existing FEC, as you have begun.  The case could also be made, for example, to eliminate FEC to address the low-latency market.  I’d prefer to let this be decided in the SG phase, but in doing so the CSDs would need to remain agnostic about the latency aspects of the applications. 

 

Regarding the meeting timing assessment, I think the March/May relationship is correct.  But I think the July/November one should instead be July/September.

 

Regards,

Paul

 

 

From: Dudek, Mike [mailto:Mike.Dudek@xxxxxxxxxx]
Sent: Tuesday, January 02, 2018 2:01 PM
To: STDS-802-3-NGMMF@LISTSERV.IEEE.ORG
Subject: Re: [802.3_NGMMF] what form for our objectives? please discuss

 

I think this brings up four different issues.    

  1.  Whether we think we should have objectives for just one or more than one solution for a given speed.   I think this should be determined in Study Group as it would affect the CSD’s.  (Robert assumed just one in the suggestions below and I think I agree with that).
  2. Whether we have objectives for both 200 and 400G.   I think this should be determined in Study Group as it would affect the CSD’s.   I think we should have a 200G objective (and the objective should be for a single fiber pair).
  3. Do we specify the number of fibers for the speed in the objectives.   I think this does affect the CSD’s and would prefer to state 4 fiber pairs for the 400G objective.
  4. Whether we will allow new FEC/Logic.   (i.e. PHY versus physical layer).  We could leave that to Working Group but I think there will be a much stronger case for “Broad Market Potential” is we use the existing FEC/Logic so I prefer having “physical layer specifications”.

 

Beth had an interesting analysis about the study group timing for the 100G per lane group that I think would apply to this group as well.  It shows that the study group either has to wrap up its work in March (with the first task force meeting in May) or wait until July (with a first task force meeting in Nov).    My suggestion is that we target  July  and try to make the hard decisions in Study Group (i.e. have the more specific objectives).

 

 

From: Lingle, Robert L (Robert) [mailto:rlingle@xxxxxxxxxxxxx]
Sent: Saturday, December 30, 2017 6:33 AM
To: STDS-802-3-NGMMF@LISTSERV.IEEE.ORG
Subject: [802.3_NGMMF] what form for our objectives? please discuss

 

Dear Colleagues,

 

Happy New Year!

 

Please consider several options for the form of objectives that will be produced by the Study Group for “Next-Gen 200 & 400 Gb/s PHYs over fewer MMF pairs than Existing Ethernet Projects & Standards.”  These differences arise from 1) different forms of objectives in recent projects plus 2) whether we should choose the number of fiber pairs for a speed in Study Group vs. Task Force. (I only included single-pair and even numbers of fiber pairs, since those are the practical choices.) Historically we have not specified that a specific grade of MMF was required to meet the objective. I show a 400 Gb/s example below, but the options would apply to a 200Gb/s objective as well.

 

Q. What are the pro’s and con’s of these options? Which do you prefer?

 

  • Define a 400Gb/s PHY for operation over fewer than 16 pairs of MMF with channel lengths up to at least 100 m.
  • Define a 400Gb/s PHY for operation over (specify 1 vs. 2 vs. 4 vs. 8) pairs of MMF with channel lengths up to at least 100 m.
  • Provide physical layer specifications which support 400Gb/s operation over fewer than 16 pairs of MMF with channel lengths up to at least 100 m.
  • Provide physical layer specifications which support 400Gb/s operation over (specify 1 vs. 2 vs. 4 vs. 8) pairs of MMF with channel lengths up to at least 100 m.

 

Let us have a healthy discussion in advance of the 1/11 telecon, so we can begin to write down draft objectives.

 

Warm regards,

Robert

 

Robert Lingle, Jr., Ph.D.
Acting Chair, IEEE 802.3 NGMMF Study Group

Director, OFS Systems & Technology Strategy

2000 Northeast Expy | Norcross, GA 30071
Office: 770-798-5015
Mobile: 404-886-3581