Last week, we attended a webinar on Optical Trends in Data Center Networks, hosted by Lightwave Online. This webinar covered some of the most important developments in data center networking technology, particularly advancements in 400G optical transceivers and beyond. Here are some of our key takeaways:
1. Leading from the front. Cloud services providers are driving the innovation of high-data-rate technology. The big four (Microsoft, Google, Amazon, and Facebook) are expected to account for nearly 100% of 400G port shipments in 2019; by 2022, even as the rest of cloud providers catch up, the big four will still make up roughly half of the total.
2. New bandwidth demands require innovation. Bandwidth demands for data centers show no signs of slowing down:
As shown above, in order to achieve these data rates (particularly over long distances), transceivers will need to evolve beyond traditional data transmission schemes. Coherent optics will become increasingly necessary in this application. For 400G, almost every IEEE standard is built around PAM-4, which uses phase-amplitude modulation to send more data per bit than traditional transceivers.
3. Pluggable optics aren’t going away. Demand for bandwidth density in the 400G space has reopened the debate of onboard vs. pluggable optics. While integrated/onboard optics have found a place in some advanced core router and supercomputing applications, pluggable optics are expected to remain the preferred option for data center networking. This is because pluggable optics facilitate network maintenance and enable network operators to pay as they grow. However, this debate is expected to emerge again in a few years with the advent of 800G, at which point pluggable optics may no longer be feasible.
4. A niche for silicon photonics (SiP)? Silicon photonics, or the integration of most transceiver functionality on a wafer-thin silicon chip, have been a hot topic in networking circles for several years. However, this technology may finally have found a niche to break through: intermediate-reach 400G applications (roughly 500m to 2km). However, multimode optics still have the edge for shorter reaches because their simpler design offers cost, power, and density advantages.
5. 400G form factors: too close to call? While we’ve predicted that the backward compatibility and port density advantages will sway the market toward QSFP-DD, this may not happen in the immediate future. OSFP and QSFP-DD will likely coexist in the market for the foreseeable future.
6. There is still work to be done. One of the gaps in the current body of standards is 400G duplex transmission over multimode fiber. Such a standard would be critical to network operators seeking to extend the life of their multimode fiber infrastructure. It is possible that SWDM (shortwave wave division multiplexing) may provide key to filling this void in the market.
Want to learn more about what’s to come? Subscribe to our blog to get the latest on industry trends.