Takai-san,
There continues to be confusion as to what cost multipliers (like 4x and 2x in
your email) are referenced to. There are at least three possibilities:
- 2008
10GE LR XFP cost, as used for all cost multipliers in cole_04_0708, i.e.
costs for all years are referenced to constant 2008 10GE LR cost
- 2010
10GE LR XFP cost, as used for all cost multipliers in traverso_02_0708, i.e.
costs for all years are referenced to constant 2010 10GE LR cost
- Decreasing
10GE LR XFP cost, as implicitly used in some reflector emails, i.e. costs
for a given year are referenced to 10GE LR cost in that same year.
Other cost references may have been used
as well.
When you state that CWDM cost floor is 4x
and Serial cost floor is 2x, are you are referencing both to decreasing 10GE LR
XFP cost, or are you referencing one to decreasing cost and the other to
constant cost?
Going forward, I would recommend that anytime
a cost multiplier is used that it is accompanied by what it is being referenced
to.
Thank you
Chris
From: Atsushi Takai
[mailto:atsushi.takai@xxxxxxxxxx]
Sent: Saturday, August 16, 2008
8:56 AM
To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3BA] Discussion
on 40G for ="" 10 km SMF
I also back from my vacation.
However I will visit my hometown for my
mother's "Hatsubon" (Japanese important ceremony).
Thus I can not discuss promptly but I would
like to try to discuss.
> I wouldn't want to predict much
difference in a future poll/vote from the 68% vs 32% we had in Denver.
It may not be true because I received
e-mails from some people who became serial supporter.
(I do not avoid the possibility that other
some people changed change to CWDM.)
I believe we have to define standard using
"right" technology as Gary
pointed.
I believe CWDM is not "right"
technology.
I believe that a standard must give some
profit or merit.
The largest profit in this standard
is to give wider bandwidth with lower cost and lower power consumption, and
maybe smaller size.
I was informed, we are requested
6x10G cost at least as a target for 100GbE (10x10G bandwidth).
Thus the "right"
standard must be achievable to lower cost/bit and lower power
consumption/bit target.
During the discussions, it is not clear
how to reduce cost less than 4x in case of CWDM.
Or it seems it is almost impossible
to reduce cost from 4x significantly.
Thus I have a big
question mark for CWDM that may not improve at least cost/bit.
I believe CWDM is not
"right" technology.
I believe serial is
"right" technology.
How long have we been using 100M, 1G
and 10G standard?
We must not decide only for the
"best" solution in 2010 or so.
The "right" standard
must remain long term, say longer than 10 years.
Many discussions were how to reduce
cost and when crossover would happen in case of serial.
Ali's mail reminded me the Newport
CMOS.
I discussed with Newport in Q1, 2000 in detail
I know the same level of 40G
CMOS IC today.
And I know there needed some step
from prototyping IC to market available IC.
Also I know many efforts for 40G DFB.
We used GPPO for 10G in early stage
but we were successful to remove it.
We know how to reduce cost for 40G
serial in near future.
Our target is 2-2.5 times of 10G.
I think MSA is enough for short
term solution if necessary.
We may establish new MSA based X40
MSA to fit for 40GbE.
I think we will take off CWDM
from the table until CWDM will be proven the significant cost/bit
improvement.
We should focus if serial is right
technology for standard, while I believe no doubt.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Atsushi Takai
Marketing Division, Opnext Japan, Inc
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
----- Original Message -----
Sent:
Friday, August 15, 2008 10:59 PM
Subject:
Re: [802.3BA] Discussion on 40G for ="" 10 km SMF
Thananya,
I'm just back from a week's vacation and just finished
reading all the reflector traffic on this topic. I was just coming to the
same conclusion as you when I finally got to your email.
Based on my assessment, we have no new information available
to us. It is already all out there and available to the group. The
difference is that people are taking that information and interpreting it in
different ways to reach their position dependent on their priorities and points
of view. I wouldn't want to predict much difference in a future poll/vote
from the 68% vs 32% we had in Denver.
We therefore have entrenched positions.
Seems the only two options are to slow the whole standard's
progress down while we debate this topic (note that we have no new information
though so not sure the value in that), or else remove the objective per
Alessandro's note.
My observation is that this is arguably gone past the
technical considerations. That is over, the majority of the data is
available. This is now essentially a non-technical issue on how we get
closure on the issue. Can the ~ 2:1 majority position get past the
75% hurdle or does the group need to make the necessary procedural
decisions to allow the standard to maintain it's agreed target schedule?
I'd prefer to debate this aspect more than further
clarifications on cost analyses...
Mark
From: Thananya
Baldwin [mailto:thananya@xxxxxxxxxxx]
Sent: Friday, August 15, 2008 3:48
AM
To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3BA] Discussion
on 40G for ="" 10 km SMF
All
-
Seeing
the traffic on this email thread, is it time to consider Alessandro's possible
scenario c) [excerpt from his email dated 7/31/8]
… My
concern now shifts to how to resolve the impasse without impacting the
standard. … c) We give up the objective (I can't believe I
am writing this:-) because we can't get consensus. At that point I think the
market will decide.
Would
be an interesting data point to see where the TF stands on this option.
--------------------------------------------
Thananya Baldwin
Director
of Strategic Programs
Ixia
thananya@xxxxxxxxxxx