[j-nsp] The Switch is ON !!!

[Richard A Steenbergen]

On Wed, Jan 30, 2008 at 12:09:07PM -0800, bill fumerola wrote:
> so i agree with everything RAS said. per usual. :)

Woo. :)

BTW, I'd like to point out one additional item which is publicly available 
information but which generally seems to have been overlooked so far. If 
you take a close look at the "coming not-very-soon" EX8200 on page 6:

http://www.juniper.net/solutions/literature/brochures/150057.pdf

You'll notice 8 distinct 10G ports per blade. At first glance one might be 
tempted to believe they are XFP ports, but those are SFP+ ports. Why is 
this a problem? Because SFP+ achieve their density not by significantly 
reducing power draw, but by eliminating the higher end power classes which 
are necessary to drive medium and long reach optics (40km ER, 80km ZR, any 
DWDM tuned optics, etc). There is a thread on exactly why this sucks so 
bad over on cisco-nsp, but the bottom line is that if you have an SFP+ 
product you will NEVER be able to do long reach optics (let alone at the 
very reasonable prices or 40-channel DWDM frequencies available in 
commodity XFP today).

I'm personally baffled by Juniper's decision here, it's not like they even 
need SFP+ to achieve the density required. On the Cisco Nexus 32-port 10G 
SFP+ blade (which also does FC), or the 48-port 10G SFP+ 1U Arastra box, 
there is a legitmate excuse for using SFP+ at the expense of long reach 
optics, but on the Juniper 8-port full-sized blade there is absolutely no 
reason Juniper should not be using XFP here.

I would encourage anyone who is interested in this product and who might 
ever want to use long-reach optics in it to talk to their account team 
about XFP instead of SFP+ blades NOW before this horribly bad idea 
progresses any further.

-- 
Richard A Steenbergen <ras at e-gerbil.net>       http://www.e-gerbil.net/ras
GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)