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

Re: Ethernet_to_SONET


These proposals are specific to how a non-traditional transmission protocol
gets mapped into the SONET payload.  The inference is that the SONET
transmission vendors will be implementing the non-traditional interfaces as
path tributaries on the SONET LTEs.  In my role at MCI, I have seen several
vendors that are doing that.  Currently, the different vendors are not
interoperable.  The proposal in 268r1 will not only allow for vendor
interoperability, but also addresses the issues of deterministic performance
over SONET transmission systems.

The clocking of 802.3 interfaces will be implemented in the LTE attached
interfaces.  The demarcation between the 802.3 clocking standard and the SONET
clocking will be in the LTE multiplexing matrix.  Given the bit width
difference between the slower speed 802.3 interfaces and the current OC192
SONET systems, I don't foresee any difficulties.

The biggest issue is the removal of the IETF inspired use of HDLC
encapsulation.  I have experimented and verified that a major performance
degradation is caused by certain forms of binary data through the use of HDLC.
Those protocols that are mapped into SONET payloads this way are subject to
non-deterministic performance issues.  This was so readily recognized that the
previous proposal for 802.3 mapping, that did use HDLC, was voluntarily
withdrawn by that author.

While there are still issues and work to be done, such as the specific use of
some of the header frame bytes, I think that this is a much better solution
than was previously proposed.  I can only present my personal views on this
subject.  Those of the 802.3 voting body that want to get involved, will be
making the decision on how to respond.

Thank you,
Roy Bynum

Tom gandy wrote:

> Well I've read Roy Bynum's email and the two WORD documents which were
> linked and downloaded from an ITU FTP site.  Interesting stuff and thanks
> for the link.  However, I am confused as I don't see any mention of the
> problems of mapping from the plesiochronous clock domains of Ethernet (up
> to +/- 100 ppm) into a Synchronous Hierarchy.
> Of course, there are also issues with Inner Packet Gap shrinkage which
> might cause some problems for the newly defined IPG/Preamble frame.
> A long time ago there was an ANSI effort started to map FDDI frames into
> SONET.  100BASE-X directly follows from the physical layer aspects of FDDI.
> Anyway, Raj Jain's "FDDI Handbook" discusses this whole clocking issue and
> the necessary adding and subtracting of bits.
> I'm probably missing something here, but as it stands, I'm confused.
> Thanks
> Tom Gandy
> Industrial Catalyst