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

Re: [STDS-802-3-400G] My competing 2-km proposals



In the second paragraph, the number two determinant of cost should have been stated as manufacturing yield. It has the virtue of being quantifiable.


Chris

 

From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx]
Sent: Monday, May 18, 2015 10:40 AM
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: Re: [STDS-802-3-400G] My competing 2-km proposals

 

The proposal to adopt 4x100G PSM4 for 500m, 4x100 WDM for 2km, and 8x50 WDM for 10km will result in three very different PMDs. The similarity between 4x100G PSM4 and 4x100G WDM is superficial, and stops at wavelength count. All the optical specification which actually define an optical interface are substantially different. WDM is not converted from PSM4; it is a separate development.

 

The number one determinant of cost is volume.  The number two determinant of cost is manufacturing margin. The number of wavelengths has never been an absolute determinant of cost. For mature technology fewer wavelengths is lower cost. For a new technology rate, the higher speed always has significantly higher cost. 4x100G WDM will be the highest cost solution, once components are available to make it manufacturable, which is several years away. There is not a single precedent to support the claim that a new technology rate at introduction is lowest cost. This was certainly not true for either 10G or 25G.

 

Further, by adopting 3 very different PMDs will splinter limited industry R&D resources into developing 2 very expensive new technologies (50GBaud) and 1 derivative technology (25GBaud) for 3 modest volume applications. Today, the industry is struggling to find sufficient R&D resources to reduce the cost of 10G and 25G optics which will continue to be the dominant volume interfaces for many years to come. The main industry focus has to remain on 10G and 25G.


Chris

 

From: Brad Booth [mailto:bbooth@xxxxxxxx]
Sent: Monday, May 18, 2015 9:56 AM
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: Re: [STDS-802-3-400G] My competing 2-km proposals

 

Actually, I prefer the 4x100G lambda solution for 2 kms. If one assumes the 8x50G lambda solution is the preferred proposal for the 10 km reach, I can understand the logic applied to see the 2 km reach version as a reach optimized solution for that market space.

 

But, if I look at the 2 km reach as it's own market where the maximum reach is 2 km on a duplex fiber, then it seems to me that this technology should more closely resemble the 500 m reach solution. If a 4x100G PSM4 solution is the preferred proposal in that space, converting that solution from PSM4 to WDM would be preferred. While there is added complexity to convert a PSM4 solution to a WDM solution, I believe that the industry would be able to capitalize on the 500 m reach volume to help drive down the cost of the 2 km WDM version.

 

One of the other reasons I favor having a 4x100G lambda solution for 2 km is that the task force easily means the distinct identity requirement for each reach solution. The 500 m, the 2 km and the 10 km are all unique and distinct.

 

Thanks,
Brad

 

 

 

 

On Mon, May 18, 2015 at 7:55 AM, Dan Dove <dan.dove@xxxxxxxxxxxxxxxxxx> wrote:

Hi Jeff,

I see things exactly the same way as you expressed it. I imagine many others also will.

Regards,

Dan Dove
Chief Consultant
Dove Networking Solutions
530-906-3683 - Mobile

On 5/18/15 10:04 AM, Jeffery Maki wrote:

Dear P802.3bs Task Force Members,

 

I wanted to explain why I am an author of competing proposals for the 2-km reach objective, namely, 4x100G-lambda (50GBAUD) PAM-4 and 8x50G-lambda (25GBAUD) PAM-4. I began with 4x100G-lambda (50GBAUD) PAM-4 because it aligned with what I believe is best from an architectural point of view, which I had reviewed in my Norfolk presentation http://www.ieee802.org/3/bs/public/14_05/maki_3bs_01a_0514.pdf. I also considered in that presentation that technical feasibility in concert with time-to-market considerations might dictate something different than architectural optimization.

 

Here are some things I also know to be true:

1.       The market needs a durable (good longevity) interoperable standard for 10 km reach. This reach is where the market demands zero issues for interconnecting systems from different system vendors and implicitly module vendors.

2.       The market will ask for a reduced performance, 2 km reach, version and thus more importantly a reduced cost option that preserves interoperation with (1) up to 2 km.

3.       IEEE 802.3 standardization enabling higher-density solutions and/or lower cost solutions for 2-km applications will not stop (1) and (2) from continuing to be demanded and procured. Only a compelling new 10-km standard would change things for (1) and (2).

 

Thus, at this time, my priority has become 8x50G-lambda (25GBAUD) PAM-4 for the 2-km reach proposal since it aligns with the what I see as the winning 10-km reach proposal, which is 8x50G-lambda (25GBAUD) PAM-4.

 

Jeff

 

 

From: John DAmbrosia [mailto:John_DAmbrosia@xxxxxxxx]
Sent: Sunday, May 17, 2015 3:14 PM
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: [STDS-802-3-400G] Webpage Update

 

All,

The webpage has been updated with all files I have received as of now.

Please see - http://www.ieee802.org/3/bs/public/15_05/index.shtml.

 

We will start promptly at 8:30am tomorrow (Monday) morning.

 

Regards,

 

John D’Ambrosia

Chair, IEEE P802.3bs 400GbE Task Force