Higher Speed Ethernet is Taking Off, Isn't it?

Wow, 40GigE is really taking off.  It is early in its life cycle, of course, and still tiny compared to1GigE and 10GigE, but LightCounting recently found it necessary to revise our previously aggressive ramp for 40GigE to even higher numbers.  In fact, our July Market Forecast Report nearly doubles our 2013 and 2014 forecasts for 40GigE. 

So 40GigE is exceeding all expectations, right?  Well, yes and no.  Long reach 40G QSFP+ modules saw the greatest percentage increase due, we believe, to Google buying significant quantities of 2km devices with lower specifications.  But one needs to look deeper into the short reach 40G QSFP+ numbers to see what is happening. 

It is difficult to track how a module is used once it is shipped.  For example, a 40GigE module can connect directly to another 40GigE module in a switch-to-switch link.  But a 40G-SR QSFP+ module can also be split four ways with a 1:4 breakout cable to connect to four 10G-SR SFP+ modules in a rack of servers.  The switch’s QSFP ports are configured as four 10GigE ports and the scheme delivers a higher density solution to the “Top of Rack” switch.  

It turns out that as data centers flatten to a two-layer architecture, the splitting of 40G modules, four ways at 10G, allows for four times the number of 10G servers to be connected in a single cluster as each QSFP+ port on the switch can serve four 10G servers.  As this trend continues, 40G module vendors will reap the rewards of accelerated volumes.  With the majority of 40G short reach modules used in this fashion, it would be incorrect to conclude that native 40GigE is seeing such a fast adoption ramp. 

What does this mean for the deployment of 100GigE?  So far, 100GigE is used only between data centers and rarely within the data center itself.  Unlike 40GigE, that has needed reaches in the popular QSFP+ form factor, 100GigE is going through a long evolution of form factors and PMDs.  What is missing from this is a cost-effective 500 meter 100G solution for data centers.  The IEEE 802.3bm task force has tried mightily to address this need with four disparate PMD proposals made and defended. 

But committees don’t always come together. Today, the task force finds itself proposing to drop the 500 meter 100G solution from the list of objectives, as none of the four finalist proposals appear capable of reaching the necessary 75% support level.  John D’Ambrosia, Chairman of the Ethernet Alliance and Chair of the IEEE 400GigE Study Group remarked; “Developing a standard within the IEEE requires industry consensus.  If there is no consensus to move things forward, then this must be viewed as a state of the industry itself.  Given the various optical technologies being explored and developed at this time, the challenge in developing consensus to drive the development of optical standards is not surprising.  However, I am confident that the Ethernet community will ultimately achieve consensus, given its relation to developing higher speeds of Ethernet.” 

But beyond the struggles of the IEEE committee, the success of 40G in the data center could actually delay the eventual shift to 100G modules.  We have 100G-SR10 modules which could be split out ten ways as even more of a density play, but this may not happen. Using 1:10 breakout cables interconnected into a leaf and spine fabric would be a bit messy. The upcoming 4x25G SR transceivers cannot be split to four slower modules as 25G is not an Ethernet rate.  Unlike the hierarchical 3-layer architectures, flat data center fabrics work well with homogeneous links which are now largely 4x10G.  It will be quite some time before leaf and spine switches are connected with mostly native 40GigE and eventually 100GigE links.  This is explained in more depth in our “40G & 100G Datacenter Interconnects” report:

Upcoming Event Participation:

The LightCounting team will be at the following industry events:

Latest Research from LightCounting:

About LightCounting Market Research

LightCounting is a leading optical communications market research company, offering semiannual market updates, forecasts, and state-of-the-industry reports based on analysis of primary research with dozens of leading module, component, and system vendors as well as service providers and other users. LightCounting is the optical communications market’s source for accurate, detailed, and relevant information necessary for doing business in today’s highly competitive environment. Privately held, LightCounting is headquartered in Eugene, Oregon. For more information, go to, or follow us on Twitter at