Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3BA] Discussion on 40G for => 10 km SMF



On Fri, 8 Aug 2008, Jesse Simsarian wrote:

I believe that we should be focused on the problem at hand of finding the best solution for 40GbE on SMF. I have heard the opinion that 40G serial will give us higher capacity, which is not the case here where one 40GbE is carried on a fiber. Of course there are WDM optical transport systems that operate at 40 Gb/s today and will continue to do so in the future. However, I do not think we should be making decisions here based on the future development of colored optical transport.

Looking at the development of 1GE:

SX/LX, then ZX, then CWDM band ZX, then DWDM band ZX, first in GBIC, then in SFP form (intermixed).

With 10GE, first there was Xenpak, then X2, then XFP, and now SFP+. We're seeing DWDM colored Xenpaks and XFPs, don't know if they're starting to become available in SFP+ form. Vendors choosing Xenpak och X2 because this is the only form factor that can fit LX4 optics hurts us now as we want to standardize on XFP (to get rid of the framer from the optical module as this means we can use the same pluggable for POS/10GE LAN/WANPHY).

So, if the aim for 40GE as "let's get it to market quickly, and people won't use it much after 2015" then by all means, CWDM solution is ok. If the aim is to create a standard that can be used in both datacenter and ISP market for a long time, 40GE serial is the way to go as this enables later industry standards for 40GE SM with longer reaches as well as WDM enablement with colored pluggable optics and external CWDM muxes. Historically, in the 5+ year timeframe after the standard has been adopted, serial solutions have prevailed and the WDM solutions have gone away.

I have little insight in all the cost benefits etc of the proposals, I'm just giving end user perspective from how I have experienced development of ethernet pluggables the past 10 years.

If serial and CWDM solutions has even close to similar cost structure initially, I think the serial solution is the way to go.

--
Mikael Abrahamsson    email: swmike@xxxxxxxxx