you are correct that the dispersion tolerance of 40G serial
is 2.5-times less than 25G serial, but maybe you would employ
40G serial anyway only in some long-haul WDM WAN where
you are using DCMs anyway .... the uncompensated reach
is still an argument because it will take a decision in what kind
of network you will employ that PHY ....
You think obviously of some 100GBase-LX4 PMD with 25G serial speed
on all four lanes. To my knowledge, LX4 PMDs are mainly used at 1310nm
where the chromatic dispersion is very small anyway. If I take typical
Lane 0 1269.0 nm -1.2 ps/nm/km -12 ps/nm after 10km SSMF
Lane 1 1300.2 nm +1.5 ps/nm/km +15 ps/nm after 10km SSMF
Lane 2 1324.7 nm +3.7 ps/nm/km +37 ps/nm after 10km SSMF
Lane 3 1355.9 nm +6.4 ps/nm/km +64 ps/nm after 10km SSMF
(2nd column center wavelength, 3rd column D, 4th column accumulated
For NRZ you could estimate the width of the optical spectrum (per lane)
be roughly 70% of the clock speed, take another factor of two for chirp
to be on the safe side, that would give you a spectral width of ~36 GHz
per 25 Gb/s lane, or something like 0.21nm FWHM @ 1310nm. That would
the accumulated ISI for Lane 3 after 10km is probably about ~13ps at 25
or 33% UI, which is probably acceptable. I think the dispersion limit
of 40km for
10 Gb/s on SSMF @ 1550nm corresponds to ~75% UI ISI, so with 33% you
should be close to the dispersion limit but still below.
With other words, I think a potential 100GBase-LX4 interface could
be used in some inter-building/campus (or other) applications with 10km
max. reach running around 1310nm. You still would not need dispersion
compensation most likely.
Myles Kimmitt wrote:
My point was that the APL should try
to be broader than an nx10G sublayer (at least architecturally) because
there are applications that could take advantage of higher speed links,
especially over time. Try to get more mileage out of the sublayer.
I wasn't proposing that 802.3 get
involved in any WAN links - it was just an example of how
economics skew in different markets.
All that you say about dispersion is
true but 25G dispersion would be 2.5 times better than 40G. Beauty is
in the eye of the beholder.
Original Message -----
Wednesday, August 09, 2006 7:56 PM
Re: [HSSG] Topics for Consideration
I agree that there is a cost to
consuming wavelengths, and they should be used wisely. But a system
that has been engineered to support 10G rates cannot necessarily step
up to higher speeds, even if optically compatible with the existing
DWDM grid and power levels. The SM fiber's dispersion limited distance
drops as the square of the bit rate. For example, for externally
modulated sources, if the dispersion limited distance is 60 km at
10Gb/s, then at 25 Gb/s it will be less than 10km. So channels that
exceed this distance would need to be dispersion compensated. Were you
thinking that the higher rates would be used only on shorter channels,
or were you thinking of dispersion compensation technologies as
work-arounds? If the former, then consider how it impacts the
objectives. If the latter, it opens up a whole set of issues.
CommScope Enterprise® Solutions
1300 East Lookout Drive
Richardson, TX 75082
I would expand on the statement:
"Carrier want to leverage their existing DWDM layer which mean
baudrate in the 9.95-12.5 Gig".
There are two layers at which compatibility with existing DWDM systems
be achieved: the electrical layer (which this statement implies) and the
photonic layer. Compatibility at the photonic layer means
with existing DWDM wavelengths, optical filters, power leveling, etc. on
the same fiber. Is is likely that 25G (+7% FEC) NRZ optical signals are
compatible with many existing 10G NRZ DWDM systems. There is a premium
using many wavelengths in these systems and a 4x25G channel might well
cheaper and more wavelength efficient (as measured by GB/s/nm)than a
Such details are probably beyond the scope of the SG but I think it is
important to architect below the MAC to allow link speeds faster than
within the APL which make sense in certain markets and will become more
economic and widespread over time.
> I have listed dilemma we are facing:
> - Implementing 100 Gig in the near term means Nx10Gig
> - Implementing 100Gig in few years the right answer might be
> - Carrier want to leverage their existing DWDM layer which mean
> baudrate in the 9.95-12.5 Gig - If LAG implemented why not allow n
to be 4?
> - Operation with different width
> - Backward compatibility XAUI, LX4 ?
> - Greatest bandwidth demands (100+Gig) are on VSR links <50 m
> the longer reach >10Km may be able to live with 4x10Gig.
> All these means we should either define some sort of scalable
> architecture or just define LAG method and do not define any PMDs!
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.10.8/414 - Release Date: 8/9/2006
Bell Labs / Lucent Technologies
Data Optical Networks Research
Crawford Hill HOH R-237
791 Holmdel-Keyport Road
Holmdel, NJ 07733, USA
fon +1 (732) 888-7086
fax +1 (732) 888-7074