Battle of the network stars
Sure, the amount of money flowing into the optical-networking business ex ceeds the gross national product of several countries. But the number of companies floating their products on this current of potential wealth grows all the time, as well (despite the best efforts of JDS Uniphase and Nortel Networks to shrink this number via acquisition). The competition in the fiber-optic field, therefore, continues to increase in both amount and ferocity.
That makes life hell for the people trying to establish standards. The folks who tried to settle on a small-form-factor connector for standardization are still licking their wounds. Based on a recent presentation I attended at the GEC2000 Conference in San Jose, CA, recently, the folks putting together the 10-Gigabit Ethernet standard have their hands similarly full.
The 10-Gigabit Ethernet standard promises to be the most complex Ethernet standard to date, if for no other reason than that this is the first version of Ethernet expected to make a significant splash in applications outside of the enterprise. As such, the IEEE 802.3ae Task Force has had to deal with the inputs of long-haul equipment suppliers, who want the standard to work within Synchronous Optical Network/Synchron ous Digital Hierarchy (SONET/SDH) infrastructures, as well as the usual assortment of local-area-network (LAN) players, who understandably want to maintain the backward compatibility that has been the hallmark of Ethernet evolution. When pressed by such strong pulls in opposing directions, the Task Force will likely end up doing what any self-respecting committee would do-accommodate both desires. There will be a version based on a wide area network (WAN) physical layer (PHY) that will accommodate such SONET features as OC-192 (10-Gbit/sec) link speed, SONET framing, and minimal path/section/line overhead processing, as well as a LAN PHY that will be based on the "usual" Ethernet transmission techniques.
The good thing from the fiber-optic point of view about the IEEE 802.3ae's deliberations is that they will focus exclusively on optical transport-a copper-based transmission method of transmitting 10 Gbits/sec may eventually be developed but not as part of the current standards process. The problem, however, is figuring out just what kind of optical transport will get the job done. As many readers are undoubtedly aware, the current discussion revolves around whether the standard should focus on serial transmission or incorporate wavelength-division multiplexing (WDM).
The Task Force has had no shortage of people offering to assist them in their deliberations. As of this past March, no fewer than 17 different proposals had been submitted on how to get 10 Gbits/sec of Ethernet traffic from one end of a fiber cable to the other. The Task Force has set itself a deadline of July to settle on which proposals to adopt as the core of the standard. Wish them luck.
Meanwhile, things don't seem to get any easier when you let industry players battle it out among themselves. Take the idea of flexible, interactive bandwidth pro vi sion between giga bit and terabit switch/ rout ers and optical-networking gear. Admittedly, this issue is technologically difficult, particularly given the fact that the systems which will perform such operations are barely off the drawing board. In light of this difficulty, the creation of a group like the Optical Internetworking Forum (OIF) to get industry on the same page and promote interoperability in multivendor networks using such equipment was a fine idea. But now we have another industry group, the Optical Domain Service Interconnect (ODSI) coalition, which has sprouted up to tackle the same issue. While these two groups are debating whatever it is they debate, Avici and Tellium got together at the recent OFC 2000 show to demonstrate just how such dynamic bandwidth provision might work.
So now we have two industry organizations, plus a few vendor mavericks, all working on the same issue. Avici and Tellium have said they hope to work with appropriate standards bodies on what is now a proprietary solution. One might wonder which standards group is "appropriate." Both?
Stephen M. Hardy
Editorial Director and