Re: IEEE 40GE & 100GE

[Robert Bonomi <bonomi () mail r-bonomi com>]

> From owner-nanog () merit edu  Thu Dec 13 16:53:32 2007
> From: Owen DeLong <owen () delong com>
> To: "Stephen Sprunk" <stephen () sprunk org>
> Subject: Re: IEEE 40GE & 100GE
> Date: Thu, 13 Dec 2007 14:50:34 -0800
> Cc: "Chris Cole" <chris.cole () finisar com>,
>         "Mikael Abrahamsson" <swmike () swm pp se>,
>         "North American Noise and Off-topic Gripes" <nanog () merit edu>
>
>
> So, assuming this translates roughly to optics being:
>
>       $1,000                   4km
>       $1,300                  10km
>       $2,600                  40km
>
> You'd rather have to pay $2,600 for all your campus links than
> $1,300 for all your LAN links?
>
> My preference would be quite different.  I'd much rather pay $1,300 for
> the LAN links than $2,600 for the Campus links.

Previously cited vendor data indicates 'under 10km' links out-number 10-40km
ones by a factor of 10x or more.  Is there a similar break-down of 'under 4km'
vs '4-10km' links?  Also, is there data on what share of links is 'close' to 
either limit figure?  e.g. 3.9km links that might require 10km optics to 
overcome other obstacles, or 4.1km links that are 'clean enough' that 4km 
optics might work.  (not looking for share where these things _are_ true, 
just the share where such things _might_ be possible because they're 'close 
to the limit')

At the price-points shown above, if 95% of the 'under 10km' links are
'under 4km', and only 10% of the total links are '10-40km', the 10 km / 40km
options are the least expensive (barely) aggregate.  if the distribution
is 90% under 4km, 5% 4-10km, 5% 10-40km,  4km/40km is less expensive (circa
15% less) in the aggregate.

If the 4-10km link share is more than around 4% of all less than 40km links,
the 10km optin for 'short haul' will bne less costly, if you limit things
to 2 options.