[c-nsp] XFP, SFP+, ???

[Jeff Bacon]

> On 20/10/2009 15:55, Jeff Bacon wrote:
> > Using an external box to condition the wave is an option but now I'm
> > paying bux for the colored optics in the box, plus the box, plus the
10G
> > to get into the 6500, which seems like a net big lose.
> 
> The advantage that this gives you is that you don't end up paying for
> coloured xenpaks or X2s (both of which are ridiculously expensive),
and you
> do end with a transmission system which is completely vendor and
> transceiver independent.

Well, yer always dependent on _some_ vendor - be it Cisco, opt-whoever,
or whoever's making your external box. I suppose if you're avoiding
XENPAK/X2 and trying to go XFP, then your XFP is theoretically portable.


(Though you could fork for a SIP400 and the 10G SPA and then be
XFP-happy.)

> "Cheap" is a fluid concept.  Are you measuring "cheap" according to
> up-front capex right now, or by 1Y / 3Y / 5Y tco or whatever your
> depreciation time period is, or by reusability if you change from one
> client transceiver type to another?

Up-front capex, 2yr TCO (though in the given case I'm not sure what else
factors into TCO besides the MRC of the fiber path and cost of rack
space). 

It feels like a crap-shoot on XFP / SFP+, with reach issues on the SFP+
that could be an issue.

>From the docs Peter kindly posted earlier, it appears Cisco is planning
to "cope" with SFP+ with the CVR-X2-SFP10G OneX X2->SFP+ adapters.
Doesn't help now b/c it's SR/copper only, but I'd be surprised if this
didn't get fleshed out.