re: "The grief caused over the long run by having more installed MMF
plant, especially as you try to move to 10G speeds and beyond, is IMHO
not worth the relatively small short term cost optimization."
A good point, which becomes particularly germane moving forward.
According to a TIA presentation I attended today, a 40G link will
require eight (8) strands of 8MMF. And the way things look now, a
100Gbps link will require twenty (20) strands of MMF. Neither is likely
to reach beyond 100 to 150 meters max. A single pair of SMF will
suffice for both and cover eminently greater distances. Until now the
tradeoffs have been a no-brainer, favoring MMF. But as the number of
strands begin to add bulk and handling issues and the price points
continue to fall for SMF at 10G and above, it will become more trying
to decide.
Incidentally, there's a couple of good pieces in the March 2010 issue
of Lightwave Mag on the subject of falling prices of Photonic
Integrated Circuits (PICs) at 10G and above, and one on advancements
being made in Planar chips as well on pp 6 and 24, respectively,' here:
[1]http://online.qmags.com/LW0310/Default.aspx?sessionID=43F4572DBED1C1
526DB2E9CBB&cid=218039&eid=14800#pg1
Frank