RE: Issues concerning 10GbE speed standards

[mdonhowe@xxxxxxxxxx]

Let's not forget there is another parallel optics option besides WDM.
Parallel optics with ribbon fiber is an attractive choice for short run lengths.
A 4+4 transceiver could use the same electrical footprint as the proposed 4-wide
WDM approach but a ribbon optical connector as in the Hippi optical standard.
To reach 200m at 2.5 Gbps you would probably need 300 Mhz.km fiber but I imagine
it would be new fiber anyway.  Alternatively, there is also the option of
leveraging off of the Hippi 12-wide work and keeping the GbE distances.

-Mark

Mark Donhowe
W.L. Gore & Associates
750 Ott's Chapel Road
Newark, DE  19713
ph: (302) 368-2575
FX: (302) 737-2819





"Chang, Edward S" <Edward.Chang@xxxxxxxxxx> on 06/29/99 02:24:32 PM

To:   "Rogers, Shawn" <s-rogers@xxxxxx>, "'Bruce LaVigne'" <bruce@xxxxxxxxxxx>
cc:   stds-802-3-hssg@xxxxxxxx (bcc: Mark N Donhowe/WLGORE)
Subject:  RE: Issues concerning 10GbE speed standards





Shawn:

I believe the serial/parallel issue is quite different from the "Different
Standard" issue you encountered as a very unpleasant experience.

The serial/parallel issue is similar to the "Single-mode fiber or Multimode
fiber" issue, which is an option of media choice for users based on
cost-effectiveness.

Let us just look at the actual connections of the serial-or-parallel link.
The only thing different is the transceiver module -- a Serial-Module, or a
Parallel-Module which may have the same footprint to interchange each other
depending on applications.

The 10GbE SERDES chip using CMOS technology, will have 4 I/O pins at the
line side.  Each is 2.5 Gbps for both applications.

(1) For Parallel Connection.

The transceiver module will have 4 input pins for 4 different wavelength
transmitter, then to a WDM.  One output pin from transceiver module carries
4 different wavelengths signals (2.5 Gbps each)into one fiber.

(2) For Serial Connection.

The transceiver module will have 4 input pins for 4x1 electronic
multiplexer, then the single output of the 4x1 multiplexer will connected
the 10 Gbps transmitter.  Same as case (1),
one output pin from the transceiver module carries one wavelength signal
(12.5 Gbps) into one fiber.

I believe they will have the same footprint for users to easily interchange
both parts.

It seems from user's point of view, the issue is more likely, whether to
choose short-wave or long-wave transceiver depending on applications.  I do
not see problem at all.

We also should appreciate how we benefit from CD laser and VCSEL
technologies which are cheap and high BW.  Without them we do not have
today's Gigabit rate network with affordable price for LAN applications.
We did the same approach "Let market Competition shape up the most
cost-effective products" by intentionally introduced several competing
technologies in both Fibre Channel and ATM standards, which were a pleasant
memory for all of us.


Ed Chang
Unisys Corporation
Edward.Chang@xxxxxxxxxx



-----Original Message-----
From: Rogers, Shawn [mailto:s-rogers@xxxxxx]
Sent: Tuesday, June 29, 1999 8:57 AM
To: 'Bruce LaVigne'
Cc: stds-802-3-hssg@xxxxxxxx
Subject: RE: Issues concerning 10GbE speed standards



As one of those who spent much of the 80's as a Token Ring bigot, I agree.
(FYI, converts make the best zealots!) Having multiple standards providing
the same or similar solution does a disservice to the market.

Shawn Rogers

-----Original Message-----
From: Bruce LaVigne [mailto:bruce@xxxxxxxxxxx]
Sent: Monday, June 28, 1999 6:21 PM
To: Colin Mick--The Mick Group
Cc: BRIAN_LEMOFF@xxxxxxxxxxxxxxxxxxxxxxxxxx; Bruce_Tolley@xxxxxxxx;
nuss@xxxxxxxxxx; drew.perkins@xxxxxxxxxxxx; pbottorf@xxxxxxxxxxxxxxxxxx;
Peter_Wang@xxxxxxxx; rabynum@xxxxxxxxxxx; stds-802-3-hssg@xxxxxxxx
Subject: Re: Issues concerning 10GbE speed standards



I agree with Colin.  It may be tough, it may be strained at times, but in
the
end, you (and your customers) will be MUCH happier that a single solution is
standardized.  For an example of where we (IEEE) wimped out, look at 100T
and
100VG.  Let the market decide... and it did.  Yes, I worked on 100VG, and
yes,
I still think it is a technically better solution, but I would have rather
worked to make 100T better than have two solutions in the market, which just
confuses customers.  You can't plug one into another, the cable specs are
different, etc. etc. etc.  In the end, both solutions were hurt because of
the
split.  For an example of where the IEEE didn't wimp out, look at
1000Base-T.
There was plenty of strain, but in the end, one solution was picked.  Bravo!
Let's learn from our mistakes and work this thing out here, not in the
market.

Bruce LaVigne



Colin Mick--The Mick Group wrote:
>
> I think multiuple competing solutions and a "let the market decide" is a
> sure recipe for disaster.
> It guarantees inoperable solutions and promotes market confusion.
>
> Making decisions among competing techincal solutions is a tough but
> necessary part of the standards process.
>
> Colin K. Mick
> The Mick Group
> 2130 Hanover St,
> Palo Alto, CA  94306
> voice: (650) 856-3666
> FAX: (650) 494-3737
> email: ckm@xxxxxxxxxxxxx
> URL: www.mickgroup.com