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

Re: [802.3_100GNGOPTX] SMF Objective



All,

There seems to be a lot of emphasis on whether we can justify an objective for a new SMF PMD.

 

I would like to point out a couple of things.

 

First, the CFI did identify how power & cost could be decreased and density increased by changing the 10x10 interface used in SMF PHY implementations to a 4x25 interface.  I would be surprised if we do not have consensus on this point.

 

Furthermore, the CFI identified as a potential area for SMF study :

Study alternate PMD technologies to determine if there is significant opportunity for additional size, power and cost reduction.

 

I really like how Mark Nowell phrased it in his November presentation regarding alternate schemes– step cost reduction from 100GBASE-LR4.  Mark pointed out that we need to solicit feedback from component space on relative costs.  I support this perspective as well.  Dan has summarized these presentations as not demonstrating a significant cost improvement over LR4.  In my mind Dan is really driving home the point he needs to as a chair, which I found out as a chair you always need to drive-  

 

As a reminder to the group, we are a contribution driven organization.   It should not be assumed by anyone that work will just get done without contributions to drive it.  Areas not moving forward will be brought to the attention of the (Study Group / Task Force) for consideration on how to address.

 

So if no work is brought forward to justify a new SMF PMD, does it matter if we lose the SMF aspect out of the project’s objectives?  I believe in a sense it does, as it seems to me that we have consensus on developing a retimed 4x25 electrical interface to reduce the power and cost while increasing the port density for SMF PHY implementations.  Dan, I think you make a valid point that without SMF in our objectives, it seems we are just justifying this effort based on the MMF PMD development effort.  However, if we shift our attention from the SMF PMD to the SMF PHY, we have an easy way to accurately represent this effort to the WG.  So I would suggest the following –

 

Develop an optional 4x25 retimed electrical interface (CAUI-4) for 100GBASE-LR4 and 100GBASE-ER4 PHYs.

 

Furthermore, a 4x25 non-retimed electrical interface might also get developed.  If an objective for a new MMF PHY (NOT PMD) were adopted, then this objective would cover that development effort.

 

Just some food for thought.

 

John

 

 

 

 

 

 

 

 

From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx]
Sent: Thursday, December 22, 2011 7:21 PM
To: STDS-802-3-100GNGOPTX@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GNGOPTX] SMF Objective

 

Dan

 

This is a very good summary of where we are, but I would frame the choice somewhat differently.

 

At this point, as you point out, there is not enough material to make a decision on SMF. Framing the choice as 1) will there be enough material by Jan. meeting to convince 75%, or 2) should we not have a SMF objective, is the same as saying let’s not have an SMF objective.

 

The choice is 1) should we invest significantly more time to study adding a 100G SMF objective, or 2) should we not have an SMF objective. I tried to capture this choice in a late submission presentation (http://www.ieee802.org/3/100GNGOPTX/public/nov11/cole_02_1111_NG100GOPTX.pdf), which incorporated inputs from a number of SG colleagues in the lobby bar after the social (perhaps this makes it highly suspect).

 

Chris

 

 

From: Daniel Dove [mailto:ddove@xxxxxxx]
Sent: Thursday, December 22, 2011 3:03 PM
To: STDS-802-3-100GNGOPTX@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GNGOPTX] SMF Objective

 

SG Participants,

 

To be very frank, I am concerned about whether we are going to be able to demonstrate the appropriate justification for a single-mode PMD.

 

We have received presentations on various alternative methods of achieving SMF distances using anything from MTP-SMF to SiPhotonics to Multi-Level modulation to mid-wave VCSELs. Those, by themselves have not demonstrated a significant cost improvement over LR4, nor do they demonstrate broad market potential (cost relates to volume and vice-versa).

 

Please think about whether this project should include an SMF objective. If you believe it should, prepare to convince >75% of the SG that it is technically feasible, cost effective (compare against an LR4 optical module of equivalent size and electronics), and thus will garner a sufficient market potential to create a SMF objective.

 

Otherwise, we are faced with a decision in January. Do we go to March Plenary with an MMF objective only (and a basis for developing the 4x25 electrical spec), or do we delay this SG another cycle? 

 

I plan to bring in a "5 Criteria" slide set based on what I believe will be a MMF objective. That slide set will be subject to revision. If we find need for changes on the MMF element, or addition of the SMF criteria, this would have to happen in January to make our necessary  March lead-times.

 

Regards,

 

Dan Dove

 

From: Daniel Dove <ddove@xxxxxxx>
Date: Wed, 23 Nov 2011 10:40:41 -0800
To: 100G Group <STDS-802-3-100GNGOPTX@xxxxxxxxxxxxxxxxx>
Subject: SMF Objective

 

Hi Folks,

 

Just a gentle reminder for all.

 

The goal of a Study Group is to develop objectives.

 

As engineers, we often get caught up in the details of how things work, and how we can do them better, etc.

 

This is OK, but we should keep our eye on the target.

 

  • Define a 100 Gb/s PHY for operation over at least Z km of SMF

 

Now, this is NOT an objective, but it could be.

 

We only have to do one thing in this Study Group related to SMF; Identify the the value of Z that meets the 5 criteria, starting with Market Potential.

 

We may find that there is no need for a SMF objective, or we may find that there is a need. 

 

To demonstrate need, we must show market potential for an alternative to 100GBASE-LR4 exists, and this is most likely demonstrated by showing a cost/power advantage can be obtained by alternative approaches. We have to assume that a 4x25G interface will reduce module size and power, and hence cost, so our relative comparison is not a CFP based LR4, but rather a 4x25G electrically interface based LR4 cost model.

 

Agreed?

 

If so, then demonstrating that a reasonable volume of nR4 (relative to LR4) can achieve a significant cost/power advantage over LR4, should be the path to securing an objective for SMF. We can assume that volumes in a data center that exceed SR4 reach will exist, and will be significant relative to 10Km applications. Of course we will want to quantify that assumption, but lets run with it.

 

I think breaking down the relative cost of the optical subsystems and fiber infrastructure (in the case of MTP arguments), then demonstrating how an alternative approach lowers these costs, should be the next step. For example, suppose we break down a 4x25G LR4 module into cost components (PCB, Mech, Tx_Optics, Rx_Optics, Si, Filters) and call that 100%. Then look alternative approaches and lay them alongside (assuming common volumes) and see where they land. If a particular approach is limited to Z Km, then look at the market potential for that reach based on distribution of links that can use it. 

 

Is this not a reasonable approach to the question? If so, then lets use it as a framework for future presentation.

 

Dan