[802.3ae] Re: Clarification of the 10GBASE-W ELTE function
Please read the specifications for the use of +/- 20 PPM clock for a
"client" signal. It is for maintenance (read "temporary") use only. A
non-bit stuffed client at +/-20 PPM still has too many clock offsets and
pointer adjustments to be stable from and "Ethernet" viewpoint. The +/- 20
PPM T1/T3 signals that become payload DS1/DS3 clients are "bit stuffed"
rate adjusted to allow them to be inserted into a PRS synchronous client
signal on a SONET/SDH ADM. 10G Ethernet WAN PHY does not do that.
Any equipment vendor or service provider that attempts to use the WAN PHY
as a "client" signal will have dissatisfied customers. It will give a bad
perception of the reliability of Ethernet over optical transmission
services. Changing the clock tolerance to +/- 20 PPM only created the
perception of interoperability, not the reality. That was the reason that
I apposed it.
At 11:21 PM 1/17/2002 -0800, C. M. Heard wrote:
>On Thu, 17 Jan 2002, Roy Bynum wrote:
> > There is a single point of failure that prevents a WIS as defined
> > by IEEE from ever being an interface on SONET ADM. The client
> > interface on a SONET ADM requires clock synchronization either
> > through "loop timing" or "PRS". Neither of these are available in
> > the WIS. This lack of synchronization to a transmission network
> > prevents the ADM from being able to multiplex the WIS signal into
> > a SONET signal.
>That conclusion is false. It is is possible to multiplex the WIS
>SPE into a higher-order SONET signal because the WIS transmit clock
>is required to meet the SONET minimum clock frequency accurracy of
>+/- 20 ppm. That is good enough to ensure that pointer justifications
>can absorb the timing differences between the WIS clock and the
>stratum clock used by a SONET ADM. It is not necessary that the WIS
>be loop timed or synchronized to a stratum clock in order to do that.
>All this is spelled out in Section 2.4 of the T1X1.5 contribution.
> > As defined, the WIS can never be a client of a SONET or SDH system.
> > It can be a "peer" on an optical network only. The T1X1.5 proposal
> > provides for a "peering" between a WIS and a SONET managed optical
> > network. It does not provide for the WIS to be an interface on a
> > SONET system.
>I don't think that the authors of T1X1.5/2002-027 would necessarily
>agree, since that contribution explicitly states that "[ ... ] the
>ELTE function is intended to be realized within a SONET/SDH NE
>(potentially on a line card)."