Mark
I guess we are not in full agreement!
I do agree early solution will be based on 10x10G, CTBI helps with Gen
1 and Gen 1.5.
Just as we transition from X2 to XFP/SFP+, 100Gig Gen2 will use 25Gig
for the module interface.
Which mean just as in case of XFP/SFP+ module if you want to support
LX4 or CX4 you need to
put a reverse Mux in the module!
As you know there is no compatibility between LX4 and LRM which both
serve the same market. I expect as
the technology clock moves forward the module interface probably will
be 4 lanes with 4 Lambdas and less, you are proposing
for sake of compatibility with a Gen 1 the line interface should be 10
Lambdas. This means using awkward reverse mux module.
We should not limit the future volume drivers for sake of Gen 1 product
as we did with 10Gig E.
Thanks,
Ali
Mark Nowell (mnowell) wrote:
Ali,
I don't think we are saying the same thing but
maybe we are not being specific enough.
I feel that the versions of the optical
interface should be kept to a minimum (one per reach) to ensure
interop. So we pick a number of wavelengths and then see evolution on
the technologies from eml to dml as technology matures. The user has
no interop issues. This is what we saw happen at 10GE LR.
I think we agree that the electrical interface
will start at 10Gx10 and when that technology matures it may also move
to a narrower faster interface. We saw this with XAUI to XFI and it is
an implementation decision which did not affect user interop. XFP LR
interops with XENPAK LR.
I see the electrical and optical interfaces as
decoupled. One is an implemenation issue, the other is a
interoperation issue.
For what it is worth, one goal of the CTBI
proposal was to simplify these transitions.
Mark
Mark
You might have misunderstood me. I was trying to say the early
implementation would be based on 10x10 with 10 Lamda,
followed by 4x25G with 4 lamda. I expect when you are using 4x25G then
you will use 1310 nm for 10Km and 1550 nm for
40Km. It is natural on the early implementation based on 10x10 to use
1310 nm for 10Km and 1550 nm for 40 Km.
I believe we are saying the same thing!
Thanks,
Ali
Mark Nowell (mnowell) wrote:
Ali,
I'm not sure I share your opinion that the early
implementations will be different from the later implementations in
terms of wavelengths. From a user perspective, this creates a big
interoperability challenge. The development effort to do a single
implementation is not something you want to throw away.
If you had said, early implementations may
differ in terms of technology (eg ML going to DML) then I would agree.
This raises no interop issues and drives down the cost curve as
technology matures.
Mark
Chris
Couple of items you have shaded gray (maybe) and white (preferred) the
logic for them is not clear to me:
- Why do you have 40 Km 1310 nm with 10x10 DML and ML white and you
have 1550 nm 40Km gray?
- I also don't understand your reasoning for not selecting 4x25G
over 10Km of 1310 nm, this will be the
natural implementation when 4x25 become available as you have for
40Km!
I expect early implementation of 100G will be based on 10x10. But when
the 4x25 becomes available then
both 10Km and 40Km could go 4x25G instead of using a more complex
mux/de-mux.
Thanks,
Ali
Chris Cole wrote:
Hi Marcus,
The 100GE Technical Feasibility Alternatives table only covers the two
proposed HSSG LAN reach objectives; 10km and 40km. There is currently no
WDM network objective in the HSSG. (In fact, a straw poll at the Dallas
meeting to not have a HSSG long haul PMD reach was supported.) So to
keep the coveted green shading we would like to encourage active
investigation efforts towards the use of serial techniques for the LAN
reach objectives.
In any case, I look forward to seeing your serial PHY contributions at
the next meeting.
Chris
-----Original Message-----
From: Marcus Duelk [mailto:duelk@alcatel-lucent.com]
Sent: Wednesday, April 04, 2007 12:06 PM
To: Chris Cole
Cc: STDS-802-3-HSSG@listserv.ieee.org
Subject: Re: [HSSG] <HSSG-FOAH> Meeting
Hi Chris,
thanks for the update. I am not speaking for a component vendor
so I don't know whether my feedback here can be considered as
a solution being "under active investigation". However, from a system
vendor's perspective I think that the serial solution (the last two rows
of your table) only makes sense for a WDM network which would
be at 1550nm and at reaches of 40km and beyond. These fields are
currently white but they are, at least from our perspective, under
active
investigation ... I will make further contributions on the serial PHY at
the next meeting(s).
Marcus
Chris Cole wrote:
During today's Fiber Optic Ad Hoc call we discussed updating the SMF
alternatives table. We simplified the table entries to three key
qualifiers that are required to enable technical feasibility. The
updated table is enclosed. Any comments on the proper assignment of
the
qualifiers will be appreciated.
We would also like to get input on the alternatives that are currently
under active investigation (green shading in the table) by HSSG
participants.
During the FOAC call, the 20G/25G - 1310nm alternatives, and 10G -
1550nm alternatives were confirmed as under active investigation.
We would like to find out if the 20G/25G - 1550nm, the 50G - 1310nm
and
100G - 1310nm alternatives (shaded green) continue to be under active
investigation.
We would also like to find out if any of the white areas are under
active investigation.
Please send an email indicating if you are investigating any of the
areas not confirmed during today's FOAC call.
The purpose is to accurately reflect which investigations are taking
place so that we can get an idea of what presentations we can expect
to
see during upcoming HSSG meetings. At this point we are not making any
selection between the alternatives.
Thank you
Chris
|