Fiber Versus Copper at 40/100G

I am reminded every now and then that fiber optics has been around in the U.S. since the early 1980s and that it has always been advertised as the end-all solution for networking. But somehow, copper seems to perpetually reinvent itself to be able to compete at least up to 100 meters. And it seems that this may happen again at 40 and 100 Gig.

The current copper spec within the IEEE 802.3ba standard is for up to 10 meters of twinax assemblies that may use active equalization in both the transmitter and receiver. But there is a movement about to develop a “call for interest” for a 40G/100G IEEE project for Category cables. This wouldn’t be the first time that the copper standards lagged the optical ones – Gigabit (1000BASE-T) and 10-Gigabit (10GBASE-T) were both developed after the initial fiber-optic ones. Indeed, there is much R&D being conducted at universities (Penn State) and industry (Nexans, Inc., The Siemon Company, Broadcom) to determine what can be done. While I am skeptical (based on lukewarm response to presentations given at recent Ethernet Alliance events) that a project will materialize within the IEEE any time soon, I would not rule it out—especially since recent simulations that were based on real cabling data have shown that at least 40G twisted-pair copper systems are feasible.

What keeps me a cynic about a 40/100G Category cabling solution is the extremely slow adoption of even 10GBASE-T. While it was 1000BASE-T that really propelled the Gigabit Ethernet market in 1999, it has been 10GBASE-SR (850nm VCSEL-based variant) that has enabled 10G market growth. And we really have yet to see 10GBASE-T take off at all, mainly due to its slow adoption in Ethernet switches based on its high power consumption. As I’ve stated in the market research report I wrote for CIR earlier this year, until this is solved (which may be a long time from now according to some chip suppliers), 10GBASE-T will continue to see very slow uptake.