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

Re: Unified PMD vs. Unified PHY


Given my earlier question about why the option of a unified PMD with
separate LAN and WAN PHYs, and the other issues of the weighting of the
options in the question, I tend to suspect the validity of the survey
results.  As you might remember, the motion that carried with overwhelming
support was for a limit of seven PMDs with only an attempt to reduce below
that, and the PMD was separated from the PCS/PMA in the motion that carried.

As it is, half of the "UniPHY" options failed against the separate LAN/WAN
PHYs option.  It means that there is a limit to the support for the
"UniPHY".   The bandwidth costs of the "UniPHY" proposals currently before
the group, may limit that support even more.

I think that the issue of the PMD carries a lot more weight than does the
PCS/PMA portion of the PHY.  Given that proposals are in place that would
allow a LAN PHY at 10.0 Gb MAC transfer rate to operate at only ~4% higher
baud than a WAN PHY, a unified PMD set could be defined, with separate LAN
and WAN PHYs.  Combine this with the need to add WAN overhead processing
that would not be needed in a LAN PHY, the "UniPHY" could be well over the
10% higher cost limit that the survey results defined.

What I do find interesting is that the survey placed a LAN only PHY just
above a WAN only PHY, both at the bottom of the option results, below a very
high cost "UniPHY".  This means that there will be WAN support in 10 Gigabit
Ethernet.  This is very different than what the "powers that be" thought at
this time last year would be happening.

I personally do not want to burden a LAN PHY, that could be implemented
inside buildings and campus environments, with the management capabilities
of a WAN PHY.  There may be situations that warrant it, but it should be a
distinct option that is available to the implementers.  I think that the LAN
only oriented people should be allowed to develop the LAN PHY as they
thought they wanted it to be last year.  This would allow the WAN PHY people
to concentrate on their issues and stop having to deal with the same
encoding "options" from the LAN people time and again.

Thank you,
Roy Bynum

----- Original Message -----
From: Bruce Tolley <btolley@xxxxxxxxx>
To: Roy Bynum <rabynum@xxxxxxxxxxxxxx>; Benjamin J. Brown
<bebrown@xxxxxxxxxxxxxxxxxx>; 802.3ae <stds-802-3-hssg@xxxxxxxx>
Sent: Tuesday, March 14, 2000 2:13 PM
Subject: Re: Unified PMD vs. Unified PHY

> At 08:33 AM 3/14/00 -0600, Roy Bynum wrote:
> >  I think that the original compromise and
> >the objectives as stated are correct, there needs to be seperate LAN and
> >PHYs.
> Roy:
> I think in the first part of your statement you hit the nail on the head.
Goal #9 (Define two families of PHYs) was a compromise that came out of
several meetings that was aided by the bridge diagram originally proposed by
Howard Frazier.
> Many folks have since expressed displeasure with the idea of a bridge but
the picture gave the members of the study group a way to understand and
bound the problem.
> The goal of two PHYs agreed upon in York was as much political statement
as it was a technical statement. There was a strong feeling up to the
meeting in York of the need to limit the problem, define the goals, and get
on with the work.
> It was always clearly stated that the goals were not written in stone and
we might come back to revise them.
> Given the basis of the findings from survey conducted by Jonathan, I
conclude that there is strong support among the members of the task force to
seriously investigate the concept of the UniPHY.
> Bruce