5:25 PM -- Here's something I hadn't thought about. Who's going to work on the Terabit Ethernet standard?
John D'Ambrosia of Dell Inc. made that point earlier this week at the Ethernet Technology Summit, during a session I was chairing. Companies are getting 1Tbit/s to work in the labs, and research on Terabit Ethernet is well underway. But a lot of people who would work on any kind of 1Tbit/s standard are still wrapped up with 100Gbit/s.
That creates a timing issue.
The bandwidth growth curves D'Ambrosia has been using suggest that 1Tbit/s transport will be in serious demand somewhere around 2015.
But Institute of Electrical and Electronics Engineers Inc. (IEEE) standards can take five years: one or two years to define the standard's goals, then a fixed three-year process toward ratification.
D'Ambrosia's title is something like Ethernet technical evangelist, but I know him more as the high-speed Ethernet guy at the IEEE. He's involved with 100Gbit/s from a lot of angles, which he listed on a slide:
- Chairing IEEE 802.3bj, the 100Gbit/s backplane task force
- Chairing an IEEE Ethernet Bandwidth Assessment committee, which is analyzing future network demand, to create a foundation for things like a terabit standard
- Chairing The Ethernet Alliance's board of directors
Not everybody is as busy as D'Ambrosia, but his point was that the 100Gbit/s cycle will tie up a lot of high-speed Ethernet experts for the next few years -- into the period when they really should start work on terabit standards.
So, that's something to think about, D'Ambrosia says. Certainly people can work on more than one thing at once, but 100Gbit/s will be a major focus for the foreseeable future.
Obviously, the industry will cope. There's always a workaround, and those high-end customers -- the ones that will want 1Tbit/s first -- are employing some of networking's smartest minds themselves. If D'Ambrosia's hunch is right, those customers will just start using Terabit Ethernet without a completed standard.
Even 100Gbit/s -- which has gone quite smoothly, from a standards perspective -- hit some user dissent that led to the 10x10 multisource agreement, an ad hoc, get-it-done-now project. Early terabit systems might develop along the same lines.
It might be messy, but with the focus 1Tbit/s is getting already, it shouldn't be as messy as the 40Gbit/s generation was. I'm willing to be optimistic here.
— Craig Matsumoto, Managing Editor, Light Reading