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

Fwd: RE: PMD discussion





>Paul, Pat, and all
>
>FC, Infiniband, and the OIF are all working on short reach.
>
>I think 10 GbE customers will have a need for connections of 20 to 50 meters.
>
>Because such links can be accomplished with interfaces being defined and 
>specified by other standards bodies, the IEEE 802.3ae task force has no 
>need to alter its objectives to meet this requirement.
>
>Bruce
>
>
>
>At 12:34 PM 6/2/00 -0400, Kolesar, Paul F (Paul) wrote:
>
>>Pat,
>>I am unclear as to what is meant by VSR. 802.3ae has two short reach
>>objectives, 100 and 300 m. In your view, is VSR space not covered by these?
>>Or is VSR sub-100m to you? Either way, I don't see the point of the
>>discussion unless it is aimed at changing the objectives.
>>Paul
>>         ----------
>>         From:  Patrick Gilliland [SMTP:pgilliland@xxxxxxxxxxx]
>>         Sent:  Thursday, June 01, 2000 9:34 PM
>>         To:  stds-802-3-hssg@xxxxxxxx
>>         Subject:  RE: PMD discussion
>>
>>
>>         Paul,
>>
>>         Brad is suggesting, among other things, the VSR
>>         solutions such as 850nm might be best addressed
>>         in a different standard.
>>
>>         I believe Rich Taborek has suggested the same and
>>         is soliciting proposals for the Fibre Channel PMD
>>         working group.  Probably the natural place for this
>>         work might be a different forum.  Otherwise, we must
>>         ask the entire membership to reexamine the objectives.
>>
>>         I see no future in delaying the standard by maintaining
>>         an 850nm voting block.  If the membership does not want
>>         to standardize VSR applications within 802.3, let's take
>>         it to a different forum where it is of primary interest.
>>
>>         Regards,
>>
>>         Pat Gilliland
>>         patgil@xxxxxxxxxxx
>>
>>         --------------------------------------------------------
>>
>>
>>         At 06:17 PM 6/1/00 -0400, you wrote:
>>         >
>>         >Brad,
>>         >
>>         >There are no PMDs in the set of 5 that do not meet at least one of
>>the
>>         >objectives. As far as the 850 nm serial PMD, I believe I made a
>>strong case
>>         >at the May interim as to why it also broadly meets the criteria.
>>80% of the
>>         >market for 10GbE will be under 300 m. A solution optimized for this
>>large a
>>         >portion of the market has broad market application regardless of
>>the number
>>         >of distance objectives it covers.
>>         >
>>         >To your point on a small subset getting 100% majority, the
>>indication of the
>>         >straw poll from the May interim is that down selecting below 5 PMDs
>>this is
>>         >going in the wrong direction to achieve consensus. The poll
>>indicated that
>>         >the 5 PMD set was favored by roughly 2 to 1 compared to the closest
>>         >alternative of 3 PMDs. Further, I believe that the 3 PMDs are not
>>the same 3
>>         >among the supporters of that choice, which subdivides the support.
>> From my
>>         >perspective an inclusive approach will work better than an
>>exclusive
>>         >approach in getting to consensus. In an inclusive approach you get
>>the PMDs
>>         >you prefer, while others also get the PMDs they prefer. If you
>>really
>>         >believe the market will be best served by some subset of the PMDs,
>>you are
>>         >free to use only those.
>>         >
>>         >Regards,
>>         >Paul Kolesar
>>         >
>>         >
>>         >       ----------
>>         >       From:  Booth, Bradley [SMTP:bradley.booth@xxxxxxxxx]
>>         >       Sent:  Thursday, June 01, 2000 5:19 PM
>>         >       To:  '802.3ae'
>>         >       Subject:  RE: PMD discussion
>>         >
>>         >
>>         >       Paul,
>>         >
>>         >       You touched on a key point.  To quote you, "The norm is
>>likely a
>>         >choice
>>         >       between a small subset that is targeted for their needs."  I
>>see
>>         >this as
>>         >       applying directly to what we need to work on.  If there is
>>something
>>         >       available from another standards body (i.e. VSR VCSELs),
>>then I
>>         >would prefer
>>         >       to leave that effort in that standards body especially if it
>>doesn't
>>         >broadly
>>         >       satisfy our criteria.  I think there is a small subset that
>>the IEEE
>>         >needs
>>         >       to standardize that we (I'm talking 100% majority) believe
>>we should
>>         >focus
>>         >       our effort on to meet our objectives while providing a small
>>subset
>>         >to
>>         >       satisfy our customer's needs.
>>         >
>>         >       Cheers,
>>         >       Brad
>>         >
>>         >                       -----Original Message-----
>>         >                       From:   Kolesar, Paul F (Paul)
>>         >[mailto:pkolesar@xxxxxxxxxx]
>>         >                       Sent:   Thursday, June 01, 2000 4:06 PM
>>         >                       To:     '802.3ae'; 'Booth, Bradley'
>>         >                       Subject:        RE: PMD discussion
>>         >
>>         >                       Brad,
>>         >
>>         >                       802.3z not only supported the installed base
>>of 62.5
>>         >um
>>         >       fiber (which has two
>>         >                       bandwidth grades), but also included 50 um
>>fiber in
>>         >two
>>         >       grades. These are a
>>         >                       400 MHz-km grade (representing the worst
>>installed
>>         >base
>>         >       grade of 50 um) and
>>         >                       a newer 500 MHz-km grade that allowed the SX
>>         >solution to
>>         >       meet the 550 m
>>         >                       distance objective. I don't think customers
>>have had
>>         >a
>>         >       difficult time
>>         >                       getting GbE technologies to work in this
>>situation.
>>         >But we
>>         >       are sensitive to
>>         >                       this issue. So recognizing the need to
>>distinguish
>>         >new MMF
>>         >       from old, Lucent
>>         >                       has made the new fiber easily identifiable.
>>New MMF
>>         >cable
>>         >       and patch panels
>>         >                       are distinctly color coded to distinguish
>>them from
>>         >other
>>         >       fiber types.
>>         >
>>         >                       I cannot predict the percentage of new
>>versus old
>>         >fiber,
>>         >       since I don't have
>>         >                       a crystal ball. But I believe it will be a
>>         >significant
>>         >       amount with
>>         >                       conversion accelerating as other fiber
>>suppliers
>>         >come on
>>         >       line. Lucent
>>         >                       already shipped hundreds of kilometers of
>>new MMF
>>         >and we are
>>         >       still ramping
>>         >                       up production. Also, I believe that
>>deployment will
>>         >tend to
>>         >       occur most
>>         >                       rapidly in those customer sites that intend
>>to use
>>         >10GbE
>>         >       equipment in the
>>         >                       near term. So the absolute percentage
>>conversion is
>>         >not the
>>         >       key indicator to
>>         >                       monitor, but rather the conversion occurring
>>in
>>         >10GbE
>>         >       customers sites.
>>         >
>>         >                       When I look at the 10 port types, I see them
>>serving
>>         >several
>>         >       types of
>>         >                       customers. I believe that it will be a rare
>>customer
>>         >that
>>         >       must make a choice
>>         >                       between all 10 types. The norm is likely a
>>choice
>>         >between a
>>         >       small subset
>>         >                       that is targeted for their needs. While most
>>of the
>>         >choices
>>         >       will not apply
>>         >                       for any one customer, all of the choices
>>have their
>>         >purpose
>>         >       in serving the
>>         >                       entire customer base. Let's not loose site
>>of the
>>         >fact that
>>         >       802.3 is
>>         >                       entering new market spaces. These new spaces
>>are
>>         >embodied in
>>         >       the 10 and 40
>>         >                       km distance objectives that far exceed the
>>scope of
>>         >building
>>         >       cabling (the
>>         >                       scope of Ethernet up to now), and PHY
>>objectives
>>         >which
>>         >       address both LAN and
>>         >                       WAN. Larger and more diverse market spaces
>>will
>>         >naturally
>>         >       need a greater
>>         >                       variety of solutions.
>>         >
>>         >                       Regards,
>>         >                       Paul Kolesar
>>         >
>>         >
>>         >