40 Gigabit Ethernet Surges Ahead

By Scott Schube, Senior Analyst and Strategist, LightCounting, LLC

100GbE transceivers, which use WDM technology to transmit 4 lanes of 25 Gbps traffic, came out at nearly 150 times the cost of the latest long-reach 10GbE transceivers.

Originally the unfavored stepchild of the higher-speed Ethernet effort, targeted only at server interconnect applications years away, 40 Gigabit Ethernet has emerged as a full-fledged stop on the Ethernet roadmap. Why and how?

After the IEEE Ethernet standards task force agreed to standardize both 40GbE and 100GbE, there were already some signs that Ethernet switch vendors would implement both solutions in order to capture smaller and/or more cost sensitive customers with smaller bandwidth needs or looking to make more incremental network upgrades.  However, the true hinge point came when early near-term price estimates for long-reach 100GbE transceivers, which use WDM technology to transmit 4 lanes of 25 Gbps traffic, came out at nearly 150 times the cost of the latest long-reach 10GbE transceivers.  Prices will of course come down as the technology matures, but for the next few years 40GbE long-reach transceivers will be the much more affordable option, even when priced in dollars per bit.  As a result, system vendors have begun to look at 40GbE as the next generation of Ethernet across the board, rather than just a solution for server interconnects.  Several system vendors have begun to architect their higher-speed Ethernet ASICs, backplanes, and line card designs around 40GbE; given product lifecycles, this likely means that even when long-reach 100GbE transceiver prices come down to an affordable level, mainstream 100GbE deployment will be crowded out and delayed by incumbent 40GbE switch products.
 
100GbE will still be deployed in the near term, in large core routers, particularly those connecting to next-generation WDM networks with 40 Gbps and 100Gbps pipes (Juniper has already announced its 100GbE offering); in megadatacenters (e.g. Google, Yahoo, Amazon) with enormous short-reach bandwidth needs between core switches; and/or by Ethernet switch vendors whose market positioning requires them to have the biggest and baddest network hardware on the block.
 
The next key area of change, confusion, and opportunity for transceiver vendors is the transceiver form factor roll out for higher-speed Ethernet.  Here too technical limitations are yielding a less than ideal solution, at least in the short term.  Short-reach 40GbE applications and short-reach 100GbE applications can be served with the very compact QSFP and CXP module form factors, respectively.  However, long-reach 40GbE and 100GbE connections currently can only fit into the much larger CFP form factor.  Will users still favor using a single line card for all reaches of higher-speed Ethernet, even if it means sacrificing faceplate density for short-reach connections?  Or will they opt to use smaller QSFP and CXP for short-reach applications and a separate CFP card for long-reach connections?
 
The 40GbE/100GbE and QSFP/CXP/CFP product mix issues will make forecasting and development investment decisions more interesting - and more difficult - for datacom transceiver vendors over the next several years.

"We’re likely to see a mix of practices and module deployment based on use case, similar to the early days of 10 Gigabit Ethernet. In many ways this will make life tougher for transceiver vendors, but it also potentially opens up opportunities to differentiate and find niches. " Says Scott Schube, Senior Analyst and Strategist at LightCounting, "In particular, in the short term smaller more nimble companies may be able to take advantage of the fragmented landscape to win business away from larger vendors if they execute and target the right customers and applications."

Learn more
www.lightcounting.com

Follow us on Twitter