RE: PMD Objectives
The model that I developed, and Chris Diminico can probably verify, has new
technology overbuilds grouping the new systems and data switches
in a specific location in the existing data room, normally in the same row,
or adjacent rows. This makes the new technology overbuild cable
requirements as 30m or less. The new data room build with a mature
technology would have the systems and data switches spread out more in the
data room, making the new data room cable requirements at about 50m or less.
At 11:21 AM 8/10/00 -0400, Kolesar, Paul F (Paul) wrote:
>A few comments and a question.
>Using the GbE link model approach, the distance capability of the link is
>reduced by connection loss only if one exceeds the power budget allocated to
>connections. We allotted 1.5 dB for connection loss in MMF links in GbE, and
>MMF proposals for 10GbE follow this same approach. This amount of power is
>sufficient to support a minimum of 2 worst case connections (0.75
>dB/connection max). Therefore the power budgets contain sufficient power to
>support at least two interconnections or one cross-connection without
>incurring a distance penalty. These interconnections could, for example,
>exist at patch panels, one at either end of a "backbone" cable, that serve
>as a connection point to equipment cables (i.e. those patch cords that plug
>into equipment on one end and the patch panel on the other end). A
>cross-connection, which uses a patch cord to join the ends of two cables,
>could exist, for example, between two equipment cables.
>The "Backbone + Equipment Cable" represents a link configuration that has
>the following topology:
>Equip ----Equip Cable---x---Backbone cable---x---Equip Cable---Equip
>Where x = a connection (in this case an interconnection at a patch panel)
>This topology is useful in larger COs to support links between bays of
>equipment, which may be on different floors or across large rooms. And, as
>you pointed out, it is certainly useful for commercial building backbone
>links, especially as "risers" within buildings.
>In your delineation of the need for 30 vs 50 m, are you saying that 30 m is
>more appropriate for new data rooms built exclusively for 10GbE and that 50
>m is needed for overbuilds of existing data rooms? Or do I have that
>backwards? If I have it backwards, please explain the rationale. If I do not
>have it backwards, then I think I understand.
> From: Roy Bynum [SMTP:rabynum@xxxxxxxxxxxxxx]
> Sent: Thursday, August 10, 2000 9:58 AM
> To: Chris Diminico; stds-802-3-hssg@xxxxxxxx
> Subject: Re: PMD Objectives
> You had sent this table to me asking for my input before you sent it
> reflector in general. I have not had time to do that before you
> out. I have some very specific issues with your table. In the
> place, as I have stated to you before, I do not believe that there
> a lot of patch panels in use except for extended distance and
> applications. In situations where patch panels are used, the
> distance is greater because of the attenuation of the connectors in
> patch panels.
> What are you referring to as far as "Backbone + Equipment Cable"?
> than a few building "riser" and campus applications, I see little
> the 300m distance objective. WAN PHY (WIS) interconnections will
> at the < 50m distance or at the 10km distance, depending on whether
> be used with DWDM and/or Optical Switching.
> I would expand the distance of all of your 30m applications to 50m.
> distinction should be between overbuilding existing data rooms with
> which is very near term and the building of data rooms exclusively
> 10GbE which is more long term.
> Thank you,
> Roy Bynum
> At 02:15 PM 8/9/00 -0400, Chris Diminico wrote:
> >As I pointed out in e-mail to Jonathan, for completeness, we need
> >state all of the relevant application spaces and associated
> >order to address the applicability of the current set of distance
> >further the PMD discussion.
> >The table attached is my attempt at characterizing the cabling
> >10 Gb/s application space. The customer premise cabling lengths
> >(by survey and reference). I am collecting information on the
> >lengths associated with
> >the central office and the data center. The lengths provided for
> >center and central office are reasonable target values which may be
> >based on further study. Any help would be
> > Chris Di Minico
> >Cable Design Technologies (CDT) Corporation
> >Director of Network Systems Technology
> >Phone: 800-422-9961 ext:333
> >e-mail: cd@xxxxxxxxxxxxxx