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

Re: AW: AW: 1550nm Serial PMD Question




Rahn,

The jitter specification has to do with multiplexing one signal payload, 
along with other signal payloads into higher bandwidth payload 
systems.  You are right, clock tolerance has very little to do with optical 
multiplexing and should not be an issue in a native DWDM service environment.

I have yet to see an explanation as to why a 100 PPM clock recovery system 
is more expensive than a 20 PPM clock recovery system.  If you use the 
argument set by the legacy SONET standards body, the receivers for 10GbE 
would be less expensive if the clock were specified at 20 PPM.  I have yet 
to hear from any of the component vendors to support that argument.  All of 
the component vendors seem to believe that the SONET/SDH version is 
slightly higher in cost.

As for optical parameters, I have yet to see anyone provide a BER 
specification at the pure optical level.  All of the SONET/SDH overhead 
bytes that are really required to provide performance monitoring at the 
electrical signal level of the optical channel are included in 10GbE.  What 
is not included are the things that have historically created 
inter-operability issues between vendors.  As for finding transponders that 
could support the specification, I sure that if you talk to someone that 
has been around SONET a long time, that finding reliable optical 
transponders for the early OC192 systems was an issue.  As many of the 
people that wrote the specification were transponder engineers, I have more 
faith in their work than I do in what a legacy SONET/SDH that wants to sell 
OC768 transmission gear has to say.

10GbE is a stand alone optical protocol.  The 10GbE WAN PHY will easily 
support transduction from the IEEE optical specification to the more 
expensive ITU specification.   It will also support the re-insertion of the 
vendor specific overhead bytes in that process.  It does not easily support 
multiplexing up to OC768 or any other 40G signaling system.

Of course, with 40G you have savings of only a 2 to 1 or 4 to 1 in 
bandwidth to equipment cost over standard SONET/SDH 10G systems.  You also 
have the issue of being able to support only half the number of 40G 
wavelengths compared with current 10G technology which gives you only twice 
the bandwidth per fiber. You also have the shorter optical amplifier 
distances with 40G, which adds to the implementation costs.  It may be that 
40G and the high end optical systems may have a much higher deployment 
costs per bandwidth than 10G.

Time will tell about the actual cost of deployment and support.

Thank you,
Roy Bynum


At 01:04 PM 5/7/01 +0200, Rahn, Juergen (Juergen) wrote:
>Hi,
>The jitter has nothing to do with optical muxing. You can always use a bad
>electrical signal in WDM environment, however to really doing some more than
>only point to point links the Ethernet (the "electrical" part as jitter and
>clock tolerance) have also take into account that you are may need to
>regenerate and for this this spec is not suited. Here it should be noted
>that for our experience the CDR modules for Ethernet clock tolerance of
>+/-100ppm are currently not less expensive but generate higher cost than the
>SONET specified version. So in this respect the Ethernet is of more
>expensive nature and I do not understand your point for the "low cost
>Ethernet".
>The second aspect are the optical parameters. The Ethernet is currently
>specified in a way that no easy verification if an receive signal is in spec
>is possible. Something like this is not in line with the goal for ensuring
>interworking and low operation cost and in the past the operators have
>insisted on simple operation as operation cost is the main part of their
>expenses. So in this respect the 10GE interface spec as currently written
>will be a cost driver. In addition there is one small issue with
>availability of components as if you look around the market you find no
>single transponder supporting the specs as written in the current draft (At
>least I searched around and did not find a single unit fulfilling that
>parameter set). There are of course a lot of marketing slides around in the
>world but on verifiable technical details there is a big gap . (May be
>fulfilling more demanding specs will lower cost but normally it is just the
>other way around.)  This is valid already for the current spec.
>For WDM interworking something in addition is needed. You where mentioning
>the ITU grid. So at least what you need is a transmitter wavelength as
>specified in the
>If you want to connect the interface as currently specked  connected to a
>WDM system you are limited to the power budget Including mux and demux as
>given by the  interface. In this case you have to note that there will be a
>penalty in addition due to Xtalk at the de- mux. This penalty however is
>difficult to be considered as the power specifications in 10 GE are not
>precisely given. This situation will be worse in case of optical amplifiers
>which are normally required and used in WDM. In case of this the optical
>specs have to be amended anyway as the current specification will lead to
>unknown optical performance when using optical amplification. If an optical
>spec that would allow WDM interworking would be developed it will not be
>different from a similar SDH spec in this respect.
>It should be noted that the only transversal compatible WDM spec is
>currently in G.959.1 while everything else on WDM is longitudinal
>compatible. So if you want to have transversal compatible optical WDM
>interface specifications you should use this G.959.1 and you are done. If
>you will do something different it is just different, not standard,  but not
>automatically cheaper.
>Regards Juergen
>
>
>
>
> > ----------
> > Von:  Roy Bynum[SMTP:rabynum@xxxxxxxxxxxxxx]
> > Gesendet:     Samstag, 5. Mai 2001 01:09
> > An:   Rahn, Juergen (Juergen); stds-802-3-hssg@xxxxxxxx; 'Marshall
> > Eisenberg'
> > Betreff:      Re: AW: 1550nm Serial PMD Question
> >
> >
> > Rahn,
> >
> > It might be noted that the jitter specification that the ITU is applying
> > is
> > to allow all signals to be multiplexed at the electrical level, not just
> > the optical level.  No one in ITU has considered an "optical multiplexing
> > only" jitter specification.  Using the "Class B" regenerator specification
> >
> > with a plezio-isosynchronous timing instead of a full isosynchronous and a
> >
> > relaxed timing specification would allow non-TDM multiplexed optical
> > services and non-"digital wrapper" DWDM wavelength services.  The larger
> > vendors are attempting to maintain their dominance and high margins for
> > transmission equipment.  I expect that ITU is very upset about 802.3ae
> > being a non-electrical multiplexable optical channel.
> >
> > Thank you,
> > Roy Bynum
> >
> > At 09:37 AM 5/4/01 +0200, Rahn, Juergen (Juergen) wrote:
> >
> > >Hi,
> > >The spec is indeed not in line to the "ITU Grid". However  it should be
> > >noted that beside this small pree condition of supporting particular
> > >wavelength there are a lot further specs required that are not at all in
> > the
> > >Ethernet document. This means this interface in the current spec is not
> > >suited for direct WDM also. For WDM you would in this case require a real
> > >SDH/SONET interface. In this respect it should also be noted that one
> > reason
> > >(beside the optical parameters that do not support effective
> > transmission)
> > >is the way the jitter is specified as this prevents the cascadability of
> > 3 R
> > >regens which are needed in WDM networks of larger dimensions.
> > >Regards Juergen Rahn
> > >
> > > > ----------
> > > > Von:  Marshall Eisenberg[SMTP:marshall@xxxxxxxxxxxxxx]
> > > > Gesendet:     Mittwoch, 2. Mai 2001 16:39
> > > > An:   stds-802-3-hssg@xxxxxxxx
> > > > Betreff:      1550nm Serial PMD Question
> > > >
> > > >
> > > > All,
> > > > A while back there were rumors that the 1550nm Serial PMD would be ITU
> > > > Grid
> > > > Spec compliant, making it DWDM 'ready'   I've downloaded draft D3.0
> > and
> > > > cannot find any references or further notations.  Can anyone shed any
> > > > light
> > > > on this or point out the relevant section in the draft?
> > > >
> > > > Regards,
> > > >
> > > > Marshall Eisenberg
> > > > Director of product marketing
> > > > Foundry Networks Inc.
> > > > 2100 Gold Street
> > > > P.O. Box 649100
> > > > San Jose, CA 95164-9100
> > > > (408) 586-1754 direct
> > > > (408) 586-1900 fax
> > > > (408) 398-0014 mobile
> > > >
> > > >
> >