Re: [EFM] Changes to 100BASE-X and 1000BASE-X PCS, 10G RS
Forwarded from Ben Brown.
Date: Mon, 01 Mar 2004 17:57:35 -0500
From: Benjamin Brown <benjbrown@comcast.net>
To: piers_dawe@agilent.com
CC: benjamin.brown@ieee.org, stds-802-3@ieee.org, stds-802-3-efm@ieee.org
Subject: Re: [EFM] Changes to 100BASE-X and 1000BASE-X PCS, 10G RS
Piers,
That is an interesting perspective. Regardless of the PAR for this
project,
we want to develop PHYs that can be used for all environments.
This is a project for the access network. The PHYs that come out of it
should be particularly tuned for that access market (and I believe they
are). If they can also be used for other markets then that's a bonus.
However, should we make these PHYs less applicable to the access
network just to make them more applicable to other networks when
the whole reason they exist is due to an access network project?
Regards,
Ben
Ben,
There is stuff in EFM that is just too good to be pigeon-holed into the relatively small and new
"Ethernet for access networks" market. EFM has done good things which the large market of general
IT/industrial/core/metro "legacy" Ethernets should be able and positively encouraged to take
advantage of, within the 802.3 of the future, and with true compliance.
In particular, the future 802.3 will allow a consistent physical infrastructure (fiber cable) that
can be upgraded 100-fold from 100 Mb/s to 1 Gb/s to 10 Gb/s, with a consistent maximum reach of 10
km on SMF. That's so much better for network operators than having to revisit link lengths, cable
types and attenuations whenever considering an upgrade. D3.1 Clause 66 presently tries to make this
hard to do by asking for currently non-compliant PCSs.
Even if the EFM group appears to be relaxed about this rules change for PCSs, 802.3 at large should
not. The optics company doesn't forbid non-standard uses of its optics, but it does want the
standard uses to be branded as standard so that IT departments everywhere can take advantage of our
work without having the special knowledge needed to read between the lines.
If you think the systems companies and users will do the right thing and carry on building ports
with legacy PCSs and 100BASE-LX10, 100BASE-BX10 and 1000BASE-LX10 optics anyway, we end up at the
same conclusion: the standard should say the right thing. Which has to be, treat 100BASE-LX10 the
same as 100BASE-FX, treat 1000BASE-LX10 the same as 1000BASE-LX.
This isn't just about names, it's about meeting objectives, recognising the good stuff, and the need
for careful review of the PCS changes. Retrospective rules changes are not good.
Piers
Piers,
Your concern appears to be over the marketing of the
GBIC/PMD-type devices.
100BASE-LX10, 100BASE-BX10 and 1000BASE-LX10 are all names of
PHYs, not optics. If an optics company can build optics that
work in both a
1000BASE-LX PHY and a 1000BASE-LX10 PHY and they adverti$e the
optics as such, then they're not wrong if a system company
misnames the PHY.
Today, common optics are built for multiple technologies (FC,
GE, SONET).
It is not the concern of the optics company how their optics are used
and what
name is applied to the port type.
The majority of the group appears to be comfortable with
requiring the
PCS to
be capable of unidirectional operation in order to use the
new EFM PHY
names.
That doesn't mean system companies can't continue to build ports with
different
or a limited set of capabilities. They just can't rightly
call them by
the new EFM
PHY names.
Ben