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

Re: Is this reflector still going??

Based on the dominance of copper in the

10 GbE CX motion defeated, 6x to 3x. Dan Dove and Ed Cady still support it.
- CX presentation
- CX survey
- Big equipment with removable transceivers must offer customers a cheap
equipment room connection.
- CX will result in the lowest cost equipment room connection, whether it will
be copper or fiber
- Fiber folks must disagree with market objective.

Dan Dove wrote:

> If so,
> I believe the reason we did not pull >75% of the vote for an objective
> at Montreal may be that we did not provide a sufficient argument for
> market demand, application space, and feasibility.

Hi Dan,

I agree with your your assertion that sufficient argument was not provided in
favor of a short-haul copper objective. Please allow me to add religion to the

1) There is a significant amount of anti-copper sentiment in the HSSG. This
argument is religious in nature and very detrimental to 802.3 standards activity
to provide the market with a range 10 Gbps Ethernet solutions at the lowest
possible cost. We have already decided that a range of 100 m to 40 km exists,
but have have voted down a 10 m distance. Since all fiber variants are generally
specified to support a 2 m minimum distance, all fiber variants can support 10 m
short-haul applications. But at what cost? I believe that the 10 Gbps Ethernet
market will suffer if competition from the copper side excluded on the basis of
religion. Given that all other PAR objectives are met, short-haul copper should
be given the chance to compete with short-haul fiber at 10 Gbps.

> I appreciate the work presented, and believe that feasibility has been
> reduced as an issue, but to really nail this one, it would be nice if we
> were to come up with more consensus internally prior to presenting to the
> We still have a chance to get this objective approved if we do the
> following;
> 1) Gather 1000BASE-CX shipment numbers from cable suppliers and prove
>    that stacking & top-of-stack solutions will use copper.
> 2) Obtain customer testimonials declaring the value of a low-cost copper
>    alternative.. I believe I can get some of these.
> 3) Demonstrate network designs that take advantage of a low-cost copper
>    link... I can do this too.
> 4) Come up with a SINGLE approach that provides low-cost connectorization,
>    cabling, and silicon and present each of those elements to HSSG.

Being an active participant in both Ethernet and Fibre Channel standards, it is
clearly evident that there is far more demand for short-haul copper as a storage
transport than a LAN transport. This is primarily due to the performance
criteria of storage applications which require low latency (i.e. short) links.
These short links are employed in a all storage equipment connections including
host to switch, switch to storage controller, switch to switch, and for FC, even
storage controller to disk drive. With the exception of backup links intended to
avoid data loss (e.g. disaster situations such as earthquakes, tornadoes,
floods, etc.), most storage links are short. Many of these short links can be
supported by cheaper short-haul copper solutions.

It is inevitable that future Ethernet and Fibre Channel PHYs will be MORE rather
than LESS common. Therefore, there clearly IS a significant PHY demand for
short-haul copper as data rates increase. Ethernet may ignore this market
demand, but the market demand itself will be met.

> Its my opinion that the optimum solution may be somewhere between the HP
> 4-up/4-down proposal and the Transcendata 1-up/1-down proposal that uses
> multi-level signalling.
> I am wondering if a 2-up/2-down scheme with 8B10B, scrambling, and
> multi-level (or bandwidth efficient 2-level) signaling may be sufficient to
> do the job. Is anyone willing to consider that and perform some analysis?

The problem with an in-between solution is that it ends up being a man-beast
with the worst of two solutions combined in a manner which may maximize
consensus, but also maximizes product cost.

In general, unless the media selected for a particular link application is
bandwidth limited, the most cost effective solution is to use a single channel.
Using multiple channels automatically engages requirements to handle skew,
mux/demux, cost of multiple Tx/Rx pairs, cost of multiple media pairs (unless
multiple subcarriers are employed), multi channel reliability loss,
multi-channel crosstalk, larger connectors supporting more pins, higher EMC,
etc. Ed Cady of FCI-Berg and I  shown in my July CX proposal that Twin-Ax has
sufficient bandwidth to support 5 GBaud @ 10 m using. Using PAM5 modulation, a
10 Gbps data rate can be achieved.

I'm concerned that any multi-channel short-haul solution, especially one which
combines more complex MAS modulation techniques, may not be cost competitive
with a short-haul fiber solution. If this is the case, there may not be a clear
reason for a 10 Gbps short-haul copper objective since the 10 Gbps short-haul
portion of this objective is already met with fiber.

I am currently working with several other HSSG members to provide architecture
proof of concept and implementation feasibility for a single channel,
PAM5-based  modulation, PHY which addresses short-haul copper applications for
10 Gbps in the same fashion as fiber applications. This is an open approach and
I welcome other HSSG members to join in this effort to make low-cost 10 Gbps
short-haul copper a reality.

That was my long-winded way of saying: I'll sign up for your #4 above!

> If we want a short-haul copper objective, we will have to nail it down by
> September to get it into the HSSG study group, otherwise, we will have to
> develop a separate study group, PAR, etc which will slow it down and make it
> much more difficult to complete in a timely manner.
> Your thoughts?
> Dan
> --
> ___________     _________________________________________________________
> _________    _/    ___________  Daniel Dove         Principal Engineer __
> _______     _/        ________  dan_dove@xxxxxx     LAN PHY Technology __
> _____      _/           ______  Hewlett-Packard Company                __
> ____      _/_/_/ _/_/_/  _____  Workgroup Networks Division            __
> ____     _/  _/ _/  _/   _____  8000 Foothills Blvd. MS 5555           __
> _____   _/  _/ _/_/_/   ______  Roseville, CA 95747-5555               __
> ______        _/      ________  Phone: 916 785 4187                    __
> _______      _/      _________  Fax  : 916 785 1815                    __
> __________  _/ __________________________________________________________


Best Regards,

Richard Taborek Sr.    Tel: 650 210 8800 x101 or 408 370 9233
Principal Architect         Fax: 650 940 1898 or 408 374 3645
Transcendata, Inc.           Email: rtaborek@xxxxxxxxxxxxxxxx
1029 Corporation Way    
Palo Alto, CA 94303-4305    Alt email: rtaborek@xxxxxxxxxxxxx