When it comes to 10G Ethernet, what's good for fiber is good for copper. Or so think two IEEE working groups readying versions of the 802.3ae standard for carrying this ultra-fast Ethernet over copper.

Proponents say 10G Ethernet over copper would be ideal for short-distance interconnections - stacking in wiring closets, switch-to-switch or server cluster links in a data center, and potentially connections to power user desktops, for example. And it would provide a less-expensive alternative to fiber for super-fast enterprise switching, proponents say.

The 10G Ethernet-over-copper standards would be complementary, with both suitable for short-distance applications but one perhaps better for high-speed connections within devices such as servers, says Dan Dove, task force chairman and principal engineer for LAN physical layer technology at HP's ProCurve Networking Business.

The IEEE's 10Base CX4 task force proposes a method for carrying 10G Ethernet over four pairs of twin-axial cabling. The distance limitation is 50 feet. In the meantime, the other IEEE group is looking at 10G Ethernet over twisted-pair wiring. This 10GBase-T method would be capable of supporting connections up to 300 feet.

Products by 2005

The CX4 group says it hopes to gain formal 802.3 committee approval for its standard early next year, and Dove says he expects to see products based on the standard shortly after that. Standardization of the 10GBase-T method would trail by at least a year.

"CX4 will help bring down costs for short data connections. Cable is more expensive than fiber, but for short-reach connections the total cost [of copper] will be lower," says Brad Booth, chairman of the 10GBase-T study group and an architect in the Intel Communications Group. "When Base-T shows up, we expect it will absorb some of the CX4 market."

The 10GBase-T study group is working on a project approval request (PAR), which must show that the technology has broad market potential, is distinct from and compatible with existing standards, and is technically and economically feasible. The study group hopes to get the PAR completed by November. If the 802.3 committee then approves the PAR, a task force can begin work on a standard. Products based on such a standard wouldn't materialize until mid to late 2005, Booth says.