40G Ethernet, or beyond Route management XML switching appliances

40G Ethernet, or beyond

So 10G Ethernet, standardized 15 months ago, is seeing but a whisper of deployment in enterprise networks, and even less than that in carrier networks. But that's not stopping industry watchers from speculating on what speed to take the venerable LAN standard to next.

In May, for example, a senior Cisco executive voiced his opinion that 40G bit/sec Ethernet could be technically feasible within two years. Doing 40G Ethernet, based on SONET technology, would be easier than doing the linear mathematical leap to 100G, said Luca Cafiero, general manager for switching, voice and storage at Cisco.

And in a recent interview, Raleigh Mann, manager of network operations at Industrial Light & Magic (ILM), said 40G is within reason for this San Rafael, Calif., movie post-production house. Coincident with a headquarters move planned for 2005, ILM is designing a network that would deliver 1G bit/sec connections to about 3,800 user machines, using multiple 10G links between each switching closet and a "very large mesh of 10G at the core for redundancy and performance." Mann anticipates close to 200 10G interconnects on the network, and says he envisions 40G within the core and to some higher-density closets.

Despite the murmurs, no one has yet initiated a "call for interest" (a formal IEEE process to ensure an idea is timely and interesting enough to pursue) on 40G, 100G or any other speed higher than 10G, says Bob Grow, chair of the IEEE 802.3 Working Group and a principal architect at Intel. "Another bump in speed has not been discussed, period," he says. "If anyone alludes to something, it's just that there is a future and that we'll eventually get around to it."

Rather, Grow says, recent calls for interest and projects have been focused on getting 10G better tuned to the enterprise market. That includes work on 10G over copper, over FDDI and, more complicated, Ethernet in the first mile (EFM), he adds.

Boosting Ethernet beyond 10G would be a big undertaking, one for which IEEE resources might not yet be available, Grow says. "We can only do so many projects and do them well. It makes sense to get [EFM] finished before undertaking another major project." With that, Grow doesn't anticipate the speed buzz getting any louder within the 802.3 working group until at least next year.