The CLR4 100G Industry Alliance, anchored by Intel Corporation and Arista Networks, has set an aggressive timetable for the development of an open, multi-vendor specification under which coarse wave-division multiplexed (CWDM) long-wavelength transceivers will be built. The transceivers will take the QSFP form factor and enable 2-kilometer 100-Gbit/sec transmission over duplex singlemode fibers. 100G CLR4 systems will consume less than 3.5 Watts, the alliance said when announcing the new technology, and it will enable the deployment of 36, 100G ports in a single RU.
When announcing the alliance and its forthcoming specification, alliance spokesmen Andy Bechtolsheim (founder, chairman and chief development officer with Arista Networks) and Mario Paniccia (Intel fellow and general manager of Intel Corporation’s silicon photonics solutions group) outlined the challenges facing today’s data centers and explained how and why CLR4 100G technology meets these challenges. Paniccia followed up the press conference with a blog post that also emphasized the alliance’s purpose and goals.
“When IEEE created 100G in 2010, there were SR10 and LR4,” Bechtolsheim explained. “Neither standard addressed needs of large data center operators, which needed the reach of multiple hundreds of meters and, most importantly, a low-power, cost-efficient design. A project that took place between 2011, 2012 and 2013 focused on next-gen optics for 100G,” he further explained. “Many proposals were promoted, but none of the cost-reduction proposals made the vote”—referring to the votes needed within the IEEE to advance a project. Therefore, no initiative moved forward within IEEE 802.3. “The objective to create a lower-cost, lower-power data-center-reach standard failed and was abandoned last summer. But the problem in the market hasn’t changed. It is more important than ever that we have a low-cost solution for 100G in the data center. The alliance has the specific goal to address this need: 2-kilometer reach for large-scale data centers with low-cost, low-power optics.”
In Paniccia’s blog post, he stated, “Data centers are becoming massive in scale, requiring longer and longer reaches for connectivity. This leaves an enormous opportunity to bring high-speed, low-power optical links that can span up to 2 kilometers in modern data centers operating at data rates up to 100 Gbits/sec. That’s more than 20 football fields … There are telecom-centric optical transceivers today operating at 100 Gbits/sec, but their power, size and costs are non-starters for the new data center. Thus, there is a huge gap that needs to be filled for reaches that span from, say, 100 meters to 2 kilometers. And that’s the problem we’re trying to address here.”
The 100GBase-SR4 specification, well on its way to finalization within the IEEE, will be a short-wave (multimode-based) “4x25” spec calling for 25 Gbits/sec per fiber over eight multimode fibers (four transmit and four receive). That specification’s reach tops out at 100 meters.
The CLR4 100G Alliance’s aggressive timetable calls for the release of a preliminary specification this month, April 2014, the quick collection of feedback on that initial spec, and the publication of a revised consensus spec in May 2014.
During a question-and-answer period during a press conference, Bechtolsheim addressed the timetable: “Nine months have passed since the failure of the spec [in the IEEE] … We can’t wait another nine months for this spec to come to fruition. We need the optics now.” He also addressed the existence of other recently formed MSA (multi-source agreement) groups making efforts similar to those of the CLR4 100G Alliance’s: “Any process that isn’t focused on time-to-market isn’t solving the problem. People want to deploy in high volume this calendar year. We’re out of time to futz around getting this done.”
