Re: Inside plant 10G fiber specs?

["Frank A. Coluccio" <frank () fttx org>]

The massive deployments of FTTH around the globe (now tens of millions of end points per quarter) have been responsible 
for driving price points for singlemode optics down considerably, and it's only going to get a lot cheaper over time 
before bottoming out. Plan to the trajectory, not the history. OM3 & OM4  MMF will provide considerably greater reach 
than their forebears, but if you're looking to centralize servers and avoid myriad intermediate TRs and LAN rooms due 
to distance constrains along the way (cascading bottlenecks), then give singlemode a closer look. You may also want to 
give IEEE 10GE PON (ratified last year) some consideration, or <forsooth!> the ITU emerging variant of 10GPON for your 
desktop and lab areas, fwiw.

BR, Frank

--- jeff-kell () utc edu wrote:

From: Jeff Kell <jeff-kell () utc edu>
To: nanog () nanog org
Subject: Inside plant 10G fiber specs?
Date: Sun, 14 Mar 2010 14:16:53 -0400

I am working up network specs for a new building, and trying to
accomodate a 10G distribution from the start.  The safe bet of running
singlemode everywhere doesn't quite fit due to cost of the optics and
the need for multimode for some other (non-network) devices anyway.

We have a legacy 62.5u/MM campus, inside and out.  The 100M to 1G
transition led to some outside plant singlemode additions due to length
restrictions on MM (even with conditioned LH optics), but the inside
plant gig was fine with multimode (typically SH optics).

10G appears to break the inside 62.5u/MM fit, with the addition that
there is no option of "LH over MM" for the little extra push beyond the
SH limit that worked with 1G optics.  Cisco's references give 10G SH
over 62.5u/MM at 26m or 33m, depending on the "modal bandwidth" of the
fiber.  At those distances it is of little benefit except some limited
vertical risers.

10G over 50u/MM looks better, depending on the "modal bandwidth" of the
fiber (66m, 82m, 300m).

So, a couple of questions...

(1) Do you have a good vendor specification (or sample cables) for
multistrand 50u/MM suitable for the 2000Mhz/km (300m) advertised reach?

(2) Inter-operability issues with legacy equipment where we have always
used 62.5?  We have at least two alarm half-duplex loops over 62.5 that
will have to "mate" with devices in this building... if I can avoid
running both types MM that would be great.

(3) Any other considerations or words of advice would be welcomed.

Thanks in advance.

Jeff