TECHNOLOGY
By KATHLEEN RICHARDS
Work is underway to develop an IEEE standard for Ethernet in subscriber access networks. The standard will address three physical layer topologies: point-to-multipoint over a single strand of singlemode fiber, point-to-point over a single fiber, and point-to-point copper over existing links.
Work on the standard began in November 2000, when the Ethernet in the First Mile (EFM) Study Group, chaired by Howard Frazier, founder and chief technology officer of Dominent Systems (San Jose, CA), started to develop the objectives that would be used to evaluate the proposed standard's technology proposals. By July 2001, when the IEEE 802 LAN/MAN Standards Committee approved the project authorization request, these objectives included support of Gigabit Ethernet fiber links of at least 10 km, Gigabit Ethernet point-to-multipoint fiber to at least 10 km, and copper links of 10 Mbits/sec or higher to at least 750 km. The IEEE 802.3 Ethernet Working Group authorized an 802.3ah EFM Task Force to draft the standard that same month, pending approval from the IEEE Standards Association Standards Board, which gave its nod to the project in October.
In November 2001, the EFM Task Force met in Austin, TX, at a conference attended by more than 200 people representing 80 to 90 companies, according to Frazier's estimates. Technical proposals ad dressing Ethernet passive optical networks (EPONs), point-to-point Gigabit Ethernet links, point-to-point copper links, and operations, administration, and maintenance (OAM) issues such as remote failure indication, remote loopback, and link monitoring were presented.
"What we are hearing at this stage is very thorough, detailed technical proposals on each of these areas," says Frazier, who now chairs the EFM Task Force. "Our expectation is that in March 2002, after having reviewed the proposals a few times, we will make a selection as to which proposals are going to make it into the standard."
The next task force meeting is scheduled for this month in Raleigh, NC. In addition to meeting the objectives, technology proposals must win 75% of the vote of those in attendance for inclusion in the baseline proposals of the standard. By November of this year, the task force should have the second draft or working group ballot, and if all goes according to schedule, the final-stage ballot in the spring of 2003. The standard is expected to be finalized in September 2003.
One point of debate at the most recent meeting in November concerned EPON topology and the maximum number of subscribers, or splits, the standard should support as well as distance.
"Currently, we have an objective to do a split ratio of at least 1:16 and a distance of 10 km," explains Frazier. "We are evaluating whether or not we can increase that to 1:32 and 20 km, deciding whether or not that is really feasible to do. We have made a great deal of progress in defining what the protocol for the operation of the PON will be-how you will actually transmit and receive data on the PON, how you'll arbitrate access from amongst the different subscribers attached to the PON."
Other topologies are more clearly defined, if not set in stone. "On the point-to-point links, I think there is general agreement that we're going to do Gigabit Ethernet. There is some interest in doing a slower speed like Fast Ethernet as well," says Frazier.
The EFM standard will also define point-to-point emulation, so that an EPON, in theory, can operate as a peer-to-peer Ethernet network. In addition to evaluating presented technology proposals, participants in the task force are working with and reviewing the decisions of other standards bodies.
One looming question is the nascent market for PON with or without Ethernet. It has yet to develop despite pocketed rollouts from incumbent local-exchange carriers (ILECs) such as SBC and small greenfield applications. "What I think is a little off about the whole thing is: number one, one of the big problems with PON in general is that businesses want point-to-point connections, and the [task force] is looking at point-to-point emulation based on what they develop for point-to-multipoint. Well, quite frankly, you can do that today quite cheaply," asserts David Gross, senior optical-networking analyst at Communications Industry Researchers (Charlottesville, VA). "The issue of course is that there is not a lot of fiber out there, but where it is, you can get very cheap service from Cogent and Yipes-whoever-just using a standard existing Fast Ethernet port."
"The telcos are really excited about PON, and they believe that they've got a market for it," counters Frazier. "We have heard the message loud and clear from them that PONs are really important to them."