March 4, 2008 -- The hot topics for this year's OFC/NFOEC reveal a generational shift in how issues are framed, according to Karen Liu, vice president of components at industry research firm Ovum RHK. The analyst says active optical cables, more holistic approaches to 10/40/100G data rates, and a new focus on size and power exemplify the new outlook. Even consumer markets appear within grasp.
"The interplay between ITU and IEEE worlds continues, but the commercial motivations have subtly shifted. When 10 Gbits/sec first came out, telecom was there first, focused on its own core capacity. Only later did the coincidence of data rates suggest that telecom vendors could leverage technology to cross over to the potentially larger datacom market," says Liu.
She continues, "The comparisons in U.S. election coverage between the baby-boomer supporters of Clinton and the millennial supporters of Obama have me thinking about generational shifts. As OFC/NFOEC approaches, technical progress is only half the story, as a shift is underway in how issues are defined."
Liu notes that active optical cables emerged suddenly as a product category in the middle of last year. "At last count, there were six vendors offering products in this area. These were not the result of a coordinated multisource agreement (MSA) development nor were they all driven by the same large customers. They don't even aim at the same application," she explains. While four vendors' products (search for Zarlink, Intel, Xloom, and Tyco) use the CX4 connector for 10-Gigabit Ethernet and Infiniband, Finisar's product is serial 10G and Luxtera's uses the QSFP connector for 40G.
The differences in thinking, according to Liu, break down as follows:
* "Old school" -- optics versus copper: Optics is clearly superior in bandwidth-distance products and therefore will win as soon as the next speed jump comes. In the meantime, it is a high-end product with a justified premium.
* "New school" -- active optical cable: The equipment maker and the end customer would rather not get dragged into the struggle between copper and optical. Universal ports and cable operations will help the total market and reduce barriers for optics. Liu says equipment makers will need to be price-competitive with copper in order to grab business and volume.
In the last year, the debate about 40G and 100G has been resolved into an agreement to go forward with both data rates with technology coordination between the two to minimize investment burden. Liu says of this notion, "Actually, there were two separate versions of the debate -- one for telecom distances and one for datacom. While telecom has moved from the expected 40 Gbits/sec to add 100 Gbits/sec to its plans, datacom has added 40 Gbits/sec to its 100-Gbit/sec plans." Proposals for both 40 and 100GbE use multiple lanes of 10 Gbits/sec to extend the relevance of 10-Gbit/sec technologies. According to Liu, the IEEE HSSG believes 100-Gbit/sec switches will enable more 10GbE in the data center.
The interplay between ITU and IEEE worlds continues, but in Liu's opinion, commercial motivations have shifted. When 10G first came out, telecom was there first, focused on its own core capacity. Only later did the coincidence of data rates suggest that telecom vendors could leverage technology to cross over to the potentially larger datacom market. Today, 40G DWDM transport is driven by core router interfaces -- which are still part of the core network. Liu says that service providers now believe that 100GbE will be "the client traffic of the future," including wholesale or business customer traffic.
"Despite interest in aligning with ITU, the datacom folks are more concerned about aligning the disparate elements of the data center: servers, switches, chips, and backplanes are in the midst of their own roadmap shifts on a number of levels," Liu contends. While part of the DWDM market continues to use fixed transmitters to minimize costs, "equipment vendors face a tough decision regarding the tradeoff between tunability and density. Until now, designs with XFP density had to sacrifice full band tunability. Now that tunable XFP is becoming available (JDSU has announced the critical high-density TOSA), some system vendors already want (fixed) SFP+ density on their platforms," she says.
In regard to high-volume applications, telecom and datacom are mounting separate attacks on the consumer or end-user market. In telecom, Liu says, FTTH deployments are driving product and practices toward fiber optics being easier to use and testable by non-specialists. Datacom vendors see the connected home moving to similar data rates as enterprise and conjecture that optics will be attractive for extending reach. "For example, HDMI (high-definition multimedia interface) requires 5 Gbits/sec in the home. Admittedly, this is a special case, but there are other examples, such as Optical Display Port and USB 3.0. That makes three separate formats for the home in the range of 5 to 10 Gbits/sec and (with optics) tens of meters," concludes Liu.
