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

Re: [STDS-802-3-400G] Proposed response to comment #558



Hi all,

This discussion illustrates why we follow a process and do things step by step.

 

First, we (usually in study group) agree on a set of project objectives and get them blessed by the working group, which we did by either unanimous or nearly unanimous support. Then we consider technical proposals for how to implement those objectives, we choose an approach, and adopt a baseline, which we also did with nearly unanimous support.

When we have produced a draft according to that baseline and are in the process of iterative review, we are tuning up the text and correcting errors, but we aren’t anymore debating whether we are trying to meet the correct objectives or whether we have adopted fundamentally the correct baseline,

 

If there is some change in the market that causes us to learn that our objectives are wrong, fair enough. Bring in a presentation to the task force, and if 75% of the participants agree, we’ll pass a motion to change our objectives.

 

If there is some technological breakthrough, for example, that causes us to discover that our adopted baseline is not the smartest thing we could do, also, fair enough. Bring in a presentation to the task force, and if 75% of the people agree, we’ll pass a motion to adopt a replacement baseline and the editors can go develop text to implement it,

 

However, comment resolution is not the place where it is appropriate to be debating whether we have chosen the right objectives or whether we have chosen fundamentally the correct baseline. Our average six-minute per comment meeting format is not the right way to make these kinds of fundamental decisions, which should be made based on evidence contained in a presentation with sufficient detail, for example, to show that there is a better technology baseline that we could have adopted, and not based on an opinion expressed in a 3-sentence comment.

Regards,

Steve

 

 

From: B Booth [mailto:bbooth@xxxxxxxx]
Sent: Saturday, September 10, 2016 9:52 PM
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: Re: [STDS-802-3-400G] Proposed response to comment #558

 

Ali,

You seem to be implying that 200G resolves the issue for the 400GBASE-FR8 in the data center market. You may be correct which would further support my point that the FR8 doesn't have BMP or economic feasibility.

Thanks,
Brad


From: Ali Ghiasi
Sent: ‎9/‎10/‎2016 12:23 PM
To: Brad Booth
Cc: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: Re: [STDS-802-3-400G] Proposed response to comment #558

Brad 

 

Looking at at the CSD broad market response attach below, I don’t see any issue with current PMD set and the responses in the CSD now that we have 200 GbE objective!

I see it more of a Deja Vu than anything else, remember when we added 40 GbE on the premise of the server market but it was widely successful in the cloud-scale data center because in 2010 the technology did not exist to shrink 100 GbE CFP module into QSFP28 till about 5 years later or two generations.  

 

This was the reason some of us push so hard in March of this year to include 200 GbE as we knew 400 GbE will not be initially a data center solution.  We need two technology evolution to fit 400 GbE into QSFP like form factor!

 

Until we achieve two technology node evolutions the 400 GbE likely will not be deployed in the data centers but we do have 200 GbE and 400 GbE is an excellent solution for internet exchanges, co-location, wireless infrastructure, service provider, operator, and video distribution.  

 

What you seem to be asking is "wanting bleeding edge technology at lowest cost and size with first generation product", that we have not yet learned how to do!  The 10 GbE took 8 years to get to SFP+, 100 GbE took 5 years to get to QSFP28, 100 GbE probably will take 10 years to get to SFP100, 10 Gbase-T took about 10 years to get to a practical power and cost, etc.

 

 

 

 

 

Thanks,

Ali Ghiasi

Ghiasi Quantum LLC

 



 

On Sep 9, 2016, at 4:26 PM, Brad Booth <bbooth@xxxxxxxx> wrote:



Dear Task Force members,

 

I'm going to be unable to attend 802.3bs on Monday and Tuesday due to some personal commitments, but I wanted an opportunity to provide my thoughts on the response to one of the comments I made against the draft.

 

In comment #558, I expressed my concern that 400GBASE-FR8 doesn't satisfy the broad market potential or economic feasibility requirements. The proposed response is as follows:

PROPOSED REJECT.

Based on data presented that supported the development of the responses to the Broad

Market Potential and Economic Feasibility Criteria, the Study Group and subsequently the

802.3 WG approved these responses. This data covered the solution that was eventually

adopted by the Task Force and is specified in P802.3bs Draft 2.0.

The SMF objective for 2km was adopted based on data presenting its need across multiple

applications. This need across multiple application areas is noted in the Broad Market

Potential Response in the IEEE P802.3bs CSD (https://mentor.ieee.org/802-ec/dcn/16/ec-

16-0057-00-ACSD-802-3bs.pdf). The commenter notes a specific implementation of

faceplate density (32 ports per 1 RU) as a requirement that must be satisfied. However,

the stated requirement is not supported by reference to an existing presentation or new

data that demonstrates this requirement across the different application areas that have

been noted in the Broad Market Potential Response.

Additionally, the commenter used the noted implementation for determining a power

envelope and cost requirements for the optical solutions, and then continues with

statements regarding "current power estimations." However, the commenter has not

provided any reference to an existing presentation or new data regarding the power

envelope, cost requirements, or "current power estimations" that can be considered.

 

In my humble opinion, the response is very weak. The Broad Market Potential and the Economic Feasibility text in the CSD does not propose a solution. The solution or PMD proposed must satisfy the CSD. Broad Market Potential requires that a solution:

At a minimum, address the following areas:

a) Broad sets of applicability.

b) Multiple vendors and numerous users.

c) Balanced Costs (LAN versus attached stations)

And Economic Feasibility requires:

Among the areas that may be addressed in the cost for performance

analysis are the following:

a) Balanced costs (infrastructure versus attached stations).

b) Known cost factors.

c) Consideration of installation costs.

d) Consideration of operational costs (e.g. energy consumption).

e) Other areas, as appropriate.

The 400GBASE-FR8 must be able to satisfy a market that is from 2 m to 2 km using duplex SMF. That reach and medium covers a very high number of data center installations which rely heavily (if not predominantly) on 1 RU switches with a specific port density and the ability to potentially change the pluggable modules. This market was supported by presentations given by multiple folks in both Study Group and Task Force meetings.

 

The response is correct that I do not provide any references to presentations for power estimates or costs. Unfortunately, I'm not in a position to release information that was shared under NDA. I do assume though that anyone "versed in the art" can perform the necessary calculations to surmise a cost and power estimate, and likewise can converse with those building or purchasing solutions that use duplex SMF for their 2 m to 2 km Ethernet connections to determine if there is a market for 400GBASE-FR8.

 

While I do not expect to change the response from a REJECT, I would like to encourage those in the Task Force to perform their own evaluations. If there is data someone can present that indicates that 400GBASE-FR8 can fit into a QSFP-DD or an OSFP, I would gladly welcome it. If not, then I'd encourage others to support this comment during the recirculation ballot.

 

Respectfully,

Brad