Re: What happened?
Robbie,
It turns out that the document that I intended to be in an attachment is
copyrighted by ANSI. It is however available to download for personal use
at: https://www.atis.org/atis/docstore/doc_display.asp?ID=167 .
As for the issue of the ELTE, there has already been an announcement from
one vendor that their DWDM system will support 10GbE. When looked at, it
can operate similar to what was envisioned for the ELTE. The real
distinction is that it was never intended for 10GbE to be a direct client
on an electrical multiplexing SONET/SDH system, only an optical
multiplexing system that used SONET or SDH as the service management
overhead. It was intended for 10GbE to function over dedicated fiber or
dedicated wavelengths, not multiplexed to other signals over the same
wavelength. If electrical multiplexing is was not intended, then +20PPM is
not needed. If it is the intention of the group for 10GbE to part of an
electrically multiplexed signal, then, and only then, does +20PPM become an
issue. Unfortunately, this is what Pat Thaler referred to as the "nose of
the camel". It is only the beginning of the slide toward the requirement
for full Stratum synchronization.
Thank you,
Roy Bynum
At 10:10 AM 5/30/01 -0700, Robbie Shergill wrote:
>Roy,
>
>Your email did not come through with T1X1 416-1999 attachment that you
>mention in the text. Can you provide this please?
>
>Also, a question to clarify some confusion in my mind: Is it intended by
>the 802.3ae that a 10GBASE-xW port will directly connect with an OC-192 port
>on a SONET box? Or is the intention that there is going to be some sort
>of "bridging" entity in the middle that I believe has been refered to as
>the ELTE? The reason I ask this is that IF there is a 10GE-to-SONET box
>in the middle, then why does a 10GBASE-xW port have to meet the +/-20ppm
>spec?
>
>-Robbie Shergill
> National Semiconductor Corp.
> 408-721-7959
>
>Roy Bynum wrote:
> >
> > What happened? Given the conversations on the reflector, I was under the
> > impression that 10GbE was going to remain at +100PPM clock. I come into the
> > meeting a little late and someone that I do not recognize is making a
> > presentation that 10GbE needed to be a full SONET client interface in order
> > to support the OTN Digital Wrapper. The next thing I knew, a vote was being
> > taken on putting the clock at +20PPM. I did not see a liaison letter from
> > either ITU or T1X1 at the 802.3 Plenary in March, so I am not sure why
> > presentations from ITU or T1X1 were being presented.
> >
> > This has me very concerned. In the first place, I did not see any reference
> > to what the standards from ITU and T1X1 require for NORMAL operation of a
> > full SONET/SDH client interface, included in the presentation that I saw.
> > In order to prevent synchronization alarms in the rest of the transmission
> > network, the SONET/SDH client is required to support loop timing or have a
> > direct link to the Stratum Clock source. The alternative operation, called
> > "Maintenance operation" allows for a +20PPM clock, which creates a sync
> > alarm which has to be manually disabled. This is the reason that the
> > telephony carriers force the Packet Over SONET vendor to support loop
> > timing and full Stratum clock synchronization link through. I know that
> > most of the people of the reflector do not understand this, which is one
> > reason that I believe that the vote to change the clock tolerance passed.
> > For those that do not have it, I have attached T1X1 416-1999. Please take a
> > look at Section 6. This is the T1X1 standard that is referenced as part of
> > the WIS (Clause 50). Note that in Clause 50, support and compliance with
> > Section 6 is specifically excluded.
> >
> > I do not believe that there is any technical reason to put the clock at
> > +20PPM instead of +100PPM. I did not see any reason given in the comments
> > that were made on draft 3.0.
> >
> > There is the argument that being forced to support +100PPM clock on the
> > regenerators will cost the service providers more. What I heard is that the
> > additional cost to the regenerators/transducers is less than 0.0001% of the
> > total cost. Most of the cost of a regenerator/transducer is in the optics,
> > which will not change. Changing the clock is in the electronics, which is
> > very in-expensive. Given that they will no longer be required to support
> > electrical level multiplexing, and the very EXPENSIVE SONET/SDH ADMs,
> > carriers will actually spend less to support 10GbE at +100PPM than
> > multiplexing it at +20PPM.
> >
> > I do not believe that this issue is settled.
> >
> > Thank you,
> > Roy Bynum
> > What happened? Given the conversations on the reflector, I was under the
> > impression that 10GbE was going to remain at +100PPM clock. I come into the
> > meeting a little late and someone that I do not recognize is making a
> > presentation that 10GbE needed to be a full SONET client interface in order
> > to support the OTN Digital Wrapper. The next thing I knew, a vote was being
> > taken on putting the clock at +20PPM. I did not see a liaison letter from
> > either ITU or T1X1 at the 802.3 Plenary in March, so I am not sure why
> > presentations from ITU or T1X1 were being presented.
> >
> > This has me very concerned. In the first place, I did not see any reference
> > to what the standards from ITU and T1X1 require for NORMAL operation of a
> > full SONET/SDH client interface, included in the presentation that I saw.
> > In order to prevent synchronization alarms in the rest of the transmission
> > network, the SONET/SDH client is required to support loop timing or have a
> > direct link to the Stratum Clock source. The alternative operation, called
> > "Maintenance operation" allows for a +20PPM clock, which creates a sync
> > alarm which has to be manually disabled. This is the reason that the
> > telephony carriers force the Packet Over SONET vendor to support loop
> > timing and full Stratum clock synchronization link through. I know that
> > most of the people of the reflector do not understand this, which is one
> > reason that I believe that the vote to change the clock tolerance passed.
> > For those that have it, take a look at T1X1.416-1999. Please look at
> > Section 6. This is the T1X1 standard that is referenced as part of the WIS
> > (Clause 50). Note that in Clause 50, support and compliance with Section 6
> > is specifically excluded, which would also exclude the +20PPM clock under
> > "maintenance operation".
> >
> > I do not believe that there is any technical reason to put the clock at
> > +20PPM instead of +100PPM. I did not see any reason given in the comments
> > that were made on draft 3.0.
> >
> > There is the argument that being forced to support +100PPM clock on the
> > regenerators will cost the service providers more. What I heard is that the
> > additional cost to the regenerators/transducers is less than 0.0001% of the
> > total cost. Most of the cost of a regenerator/transducer is in the optics,
> > which will not change. Changing the clock is in the electronics, which is
> > very in-expensive. Given that they will no longer be required to support
> > electrical level multiplexing, and the very EXPENSIVE SONET/SDH ADMs,
> > carriers will actually spend less to support 10GbE at +100PPM than
> > multiplexing it at +20PPM.
> >
> > I do not believe that this issue is settled.
> >
> > Thank you,
> > Roy Bynum