JOHN WILLIAMSON 400GNETWORKING
two generations away at least. “For that remaining population, retaining these flexibilities that enable pay-as-you-grow, post deployment interface type flexibility, mixing and matching di erent interfaces, and the ability to adopt new interfaces as they are developed, is expected to be highly valued for the foreseeable future,” he concludes. FUTURE’S BRIGHT There are many views about what the future of 400G and beyond networking might look like. Included are elements such as: the transition to more DSP- based opto-electronic platforms; more M&A activity in the optical industry as we march toward Terabit; AI as an end application playing a bigger role in driving optical interconnect solutions; flex spectrum-capable networks increasing as a percentage of deployed networks; the greater use of alien wavelengths; and the emergence of optimised superchannel transceivers. Either way, and notwithstanding some significant challenges that need to be addressed, that future looks bright.
with 400G pluggables, he thinks next generation 25.6T 1RU front panels can benefit with 800G OBO type solutions as switch ICs move to 100G serial electrical I/O. “Significant power can be saved by placing optics closer to the switch IC,” comments Sheth. “Beyond that, OBO has a lot of density and power advantages to o er as long as reliability and manufacturing concerns can be addressed.” Collings thinks that, given the highly attractive flexibilities that faceplate pluggable transceivers o er, it is expected that they will continue to be dominantly utilised in nominally all interconnect applications until a tipping point is reached where implementing non-pluggable transceivers clearly provides greater benefit in terms of capability, cost, power dissipation, or density. He allows that there are some use cases, particularly hyperscale internal DCI, which may value these flexibilities less than the remaining population of applications, but expects any non- pluggable tipping point to be one or
rates, there is some industry debate about the future roles and areas of application of coherent transmission and direct detection technologies. Sheth sees di erent roles for the two at di erent distances. He reckons direct detect is still expected to dominate optical solutions at 400G and 800G for up to 10 km, and that coherent is expected to dominate optical solutions at 400G and 800G for 80 km and beyond. “Between 10 to 80 km there is expected to be some overlap at 400G,” he suggests. “At speeds greater than 800G, coherent is expected to play a bigger role even at less than 10 km.” According to Collings, there is a very significant performance, capability and cost di erence between direct detect and coherent solutions. Given this, where a direct detect approach is serviceable, coherent-based solutions are not o ered. He believes that this gap isn’t likely to fully close in the near future and therefore, where direct detect solutions are acceptable, they will continue to dominate over coherent options. “That said, direct detect implementations do have some reasonably fundamental limitations and therefore cannot address many higher capacity applications where coherent dominates,” he accepts. In the context of 400G and higher networking, there is some discussion about the future facing pluggable transceivers and embedded OnBoard Optics (OBO). Conversations range around the issues of whether pluggables can continue to be extensively used as data rates continue to climb, and where and when onboard optics will start to play a major role. Fischer describes how, when moving to 100G/200G wavelengths, the use of pluggables on the line side has been widely accepted, and that one of the benefits is a very active supplier ecosystem driving down the cost of the optical components. It may be a di erent story at higher speeds. “Modulation speeds and formats bringing us to 600G per wavelength are extremely more demanding than current interface technologies,” he says. “There are well founded doubts in the industry if this could be done with pluggables. Also there is no ecosystem as of now.” Fischer reckons that one interesting development for pluggable line optics will be 400G-ZR and its vendor-specific extensions. “However, this will not cover all or even the majority of transport applications due to reach restrictions,” he argues. Sheth likewise sees embedded OBO having an important part to play as front panel densities move to 800G and beyond. While current 1RU front panel density of 12.8 T can be serviced PLUGGABLES AND ONBOARD OPTICS
SILICON PHOTONICS MOVES CENTRE STAGE
Silicon photonics is viewed as part of the answer to relieving potential congestion in optical networks, and as a facilitator of the move to higher bandwidths. According to the market research and strategy consulting company Yole Développement, silicon photonics is still a small market today, with sales at die level estimated to be US$30 million in 2016. However, silicon photonics has big promise, with a 2025 market value of US$560 million at chip level and almost US$4 billion at transceiver level. In a January report – Silicon Photonics 2018 – analyst Yole says the strongest demand is for 400G and that, in parallel, 200G could be only an intermediate step between 100G and 400G. “The next evolution is to develop a 400G optical port over a single fibre across 500 metres at less than $1 per Gigabit and with power of under 5 mW/Gb”, explains Dr. Eric Mounier, Senior Technology and Market Analyst at Yole. “We believe we are only at the very beginning as there is massive ongoing development worldwide for further integration,” adds Dr. Mounier. “The recent involvement of large integrated circuit foundries, such as TSMC’s relationship with Luxtera, and GlobalFoundries with Ayar Labs, are very encouraging signs showing the big promise for silicon photonics.”
SiPh vs. optical transceiver market, in US$M
$16 000
$14 000
$12 000
$10 000
$8 000
$6 000
$4 000
$2 000
2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 CAGR $0
Optical transceivers SiPh transceivers
Yole’s silicon photonics transceiver forecast
www.opticalconnectionsnews.com
17
ISSUE 12 | Q1 2018
Made with FlippingBook - Online magazine maker