TECHNOLOGY
By CAROLYN MATHAS
The first multivendor component-level interoperability tests to validate serial 850-nm optical links supporting the emerging IEEE P802.3ae standard were recently conducted. The goal of the demonstrations was to show industry advancements in 10-Gigabit Ethernet (10-GbE) serial technologies and take significant steps toward validating the feasibility of shortwave serial optics for metropolitan and access applications.
The 10-GbE 10Base-S (850-nm) Group conducting the demonstrations comprises companies developing technologies that support 850-nm serial deployment into short-reach, high-volume network applications. Chris Simoneaux, director of product marketing at Picolight Inc. (Boulder, CO), estimates that more than 50% of future 10-GbE links will be less than 100 m.
Tests were designed to prove interoperability of diverse 850-nm technologies over various types of fiber. In the past, optics designed for transport backbones dominated the 10-Gbit/sec market. In the short reach, however, transport optical module technology is costly, not based on volume manufacturing processes, and provides limited deployment of bandwidth in the short-reach point-of-presence network bottleneck, says Simoneaux. These factors make serial 850-nm optics a potentially attractive alternative.
Optical components tested in the interoperability demonstration by Cielo Communications (Broomfield, CO) and Picolight were 10-Gbit Sixteen Bit Interface (XSBI) modules operating at 10 Gbits/sec. The module footprint, although not specified by IEEE P802.3ae, is defined under an industry multisource agreement. The demonstration link consisted of data transmitted between Picolight and Cielo 10-GbE 850-nm serial transponders over more than 15 optical cables from Lucent Technologies, Corning Inc., and CDT/Optical Systems Div. Fiber lengths ranged from 26 m of low-bandwidth 62.5/125-micron multimode fiber to 450 m of high-bandwidth 50/125-micron multimode fiber. The demonstration included new enhanced bandwidth multimode fiber (EMMF) as well.
Test results demonstrated that optical data links met required bit-error rates of at least 1x10-12-less than one error for every trillion bits. By achieving a 450-m link, the interoperability group met and exceeded the criteria set by 802.3ae specifications by a 50% margin, demonstrating that the 850-nm optical link maintains integrity without degradation in performance. Finally, tests run on legacy fiber support the group's assertion that serial 850-nm enables evolutionary growth to 10 Gbits/sec for existing networks.
Jonathan Thatcher, IEEE P802.3ae task force chair and principal engineer for World Wide Packets (Veradale, WA) says, "With this demonstration, these companies have shown strong evidence of the feasibility of serial 850-nm solutions for the emerging 10-Gigabit Ethernet standard."
Given that 850-nm technology has been the most controversial as to whether it should be included in the standard, the jury may still be out. A first vote, scheduled on Oct. 15 in Los Angeles, was to indicate whether the interoperability experiment is sufficient to ensure a 10-Gbit serial 850-nm position in the 802.3ae standard. If necessary, additional experimentation or further substantiation of data may precede a potential second vote at a plenary meeting scheduled for this month in Boston.
Jeff Bisberg, product marketing manager for Cielo, believes it will succeed. "There has been doubt in the past as to whether serial 850-nm technology could perform at 10 Gbits," he admits. "Tests now prove it can. We ran tests over legacy fibers proving that the 850 enables some evolutionary growth for existing networks. That's what the experiment was trying to establish-was that this could not just be accomplished by using the latest and greatest type of fiber. We went through a number of scenarios to verify serial 850-nm is a viable technology, and we did."
According to Picolight's Simoneaux, tests such as these will help determine the future topology of how data will move through the network. It will either move through a SONET structure or packets will rule. Although both camps agree that ensuring the economical availability of high-bandwidth Internet access is the goal, the Ethernet and SONET communities view the tests, and future topology, differently.
While Ethernet proponents attempt to create a standard encompassing all elements necessary for a large-scale network competitive with SONET on a metro and WAN basis, the SONET community views serial 850-nm as an enterprise solution relegated to inside buildings where 10-GbE networks will distribute data, according to Bisberg. SONET thus will rule in the metro and WAN, with serial 850 nm relegated to enterprise access, says Bisberg in describing the feelings of the SONET camp.
"The 10-Gbit market is slow to build," asserts Bisberg. "We would like to see 10-Gbit, 850-nm technologies further expand into the SONET marketplace. With this demonstration, we reaffirmed its place in the Ethernet network. Demonstrations as early as the November time frame may include SONET demos."
Simoneaux says that the demonstration with Cielo was one of two in which Picolight recently participated. The second, a system-level technology demonstration with Foundry and Spirent, attempted to demonstrate the feasibility of serial optical systems using both XSBI and 10-Gbit Attachment Unit Interface (XAUI) architectures.
In the demonstration, an XAUI-based and XSBI-based serial optical module were plugged into either end of a system link, transmitting IEEE data traffic between a Foundry Big Iron and Spirent SmartBits box. The result of the demonstration was to show industry advancements in 10-Gbit serial technology, taking significant steps forward in validating technology feasibility of short-wave serial optics.
Other groups in the optical community also conducting feasibility tests include a 1310-nm faction testing a 10-Gbit transponder, a 1550-nm group, and a 1310-nm WDM effort. To date, only the 850-nm interoperability group and XAUI have announced results.
If you have an idea for a news story or would like to contribute to Lightwave, please contact editorial director Stephen Hardy at tel: (603) 891-9454 or e-mail: [email protected]. Ideas must focus on topics related to fiber optics and optoelectronics in the communications industry.