Two proposals considered in 10 Gig study group

Today I’d like to check in on the progress of the IEEE’s 10GBase-T Study Group, which is looking at options for running 10 Gigabit Ethernet over copper wiring.

Unlike the task force pushing the short-distance twinax cabling, 10GBase-T is not moving at warp speed. However, it is tackling a more complex problem and hopes to develop a specification that will work on the more commonly used twisted-pair wiring.

I recently spoke with the chair of the study group, Intel’s Brad Booth. He acknowledges the group is “really pushing the edge of technology.” In fact, there is not yet a working demo of the proposals on the table – only models and simulations.

Booth says there are currently two primary proposals, and both vary, depending on what grade of cabling is in use.

The first says to preserve the distance at the expense of the data rate. So the traditional 100 meters would remain constant. But on Category-5E cabling, you would really only get about 2.5G bit/sec; on Cat-6, you’d see 5G bit/sec; and only on Cat-7 would you have a true 10G bit/sec link.

The second proposal favors speed at the expense of distance. So you would get 10G bit/sec on all three grades of cabling – but distances would vary. You might be able to run 10 Gig over 40 m to 50 m using Cat-5E; 50 m to 70 m using Cat-6; and a full 100 m on Cat-7.

Big obstacles to running such a high data rate on the lower-grade cabling are the frequency supported (625 MHz is necessary) and alien crosstalk. Alien crosstalk happens when cables in a bundle generate so much noise that they interfere with each other’s signals.

By November, the study group hopes to have decided on one approach. It will then submit that to the larger bodies within the IEEE. If it approves, the group will then become a task force and can start working on a standard. Booth says it could take at least six months for a first draft and maybe two years for standardization.