With data center bandwidth growing rapidly, high-performance pluggable modules have become an important tool for network operators to scale their networks cost-efficiently. Acacia has worked closely with network operators to drive the industry’s first interoperable Probabilistic Constellation Shaped (PCS) interfaces. Acacia’s newest portfolio of silicon-based 800G coherent pluggables has been designed to double the connectivity speed from 400G to 800G to support data center interconnect (DCI) upgrades, taking advantage of next-generation routers with 800G I/O port speeds. When plugged into these routers, this family of 800G pluggables can replace traditional transport equipment across a greater range of infrastructure to meet demand for cloud and AI.

OIF Compliant 800ZR Modules
OIF compliant 400ZR has been a great success for the coherent pluggable industry with multiple suppliers and a tremendous volume of 400ZR QSFP-DD and OSFP modules deployed in metro DCI applications. With the switch and router 800G ports available, 800ZR will further reduce the cost, power, and space per 100G in the same applications. Acacia’s 800ZR pluggables support QSFP-DD and OSFP form factors fully compliant to OIF 800ZR with transmit power variants.

Acacia Delphi DSP-based coherent pluggable modules

800G ZR+ Modules with Interoperable PCS
Acacia’s 800G ZR+ modules are designed to support the recently released OpenROADM specifications that include interoperable PCS transmission capability. The 131Gbaud PCS provides an additional OSNR margin at 800G for longer reach suitable for regional DCI applications, linking multiple fiber spans and in-line amplification between data centers. Acacia 800G ZR+ pluggables support QSFP-DD and OSFP form factors with >1dBm transmit output power.

“The success of ZR+ at 400G was largely driven by its performance and interoperability, which enabled a multi-source environment to emerge,” said Scott Wilkinson, Lead Analyst for Optical Components at Cignal AI. “The interoperability now being proposed with 800G – not only in short distance applications (800ZR), but also in long distance (800G ZR+​) – expands the environment even further. Interoperable PCS will take the market for 800G pluggables beyond simple DCI into regional and even long-haul networks.”

400G QSFP-DD Module for Ultra Long-haul
Expanding Bright 400ZR+ module applications, Acacia has also added a new QSFP-DD for 400G ultra long-haul networks that include 400G QPSK and PCS with various baud rates to fit into different ROADM optical line systems. It is capable of >1dBm transmit power and high transmit OSNR over the entire C-band thanks to the integrated tunable optical filter for amplified spontaneous emission reduction.

Backed by Industry Leading Coherent Technology
Both the 800G pluggable portfolio and 400G ultra long haul pluggable are powered by Delphi, Acacia’s 9^th generation Digital Signal Processor (DSP) ASIC. Acacia boasts the broadest field-proven 400G MSA coherent pluggable portfolio in the industry with more than 250,000 ports shipped based on the Greylock DSP, including more than 10,000 Bright 400ZR+ ports. Acacia’s 800G and 400G pluggable portfolios benefit from Acacia’s 3D Siliconization approach, which applies integration and 3D stacking techniques to enable a single packaged device that includes all the high-speed optoelectronic functions necessary for coherent communications to provide benefits in cost, reliability, power, and size. These devices are manufactured using standard electronics packaging processes and result in improved signal integrity and performance through the reduction of electrical interconnects.

The Delphi DSP-based pluggable modules are planned to be available beginning in Q2 CY2024.

Additional Resources:

Blog: Key Challenges and Requirements for 800G MSA Pluggables

As generational gains in spectral efficiency, size and power become less significant, increasing baud rate is proving to be effective at significantly decreasing power per bit.

Performance Matters for Verizon as They Plan to Use Both Embedded and Pluggable Optics
Glenn Wellbrock, Director Optical Transport Planning for Verizon, talked about pluggables versus performance optimized (a.k.a. embedded) coherent modules, and how it’s important to leverage the volume that the hyperscalers are using to drive lower cost and power consumption. Glenn commented that the performance of the pluggable modules they’ve tested has exceeded their expectations. While they still plan to use performance optimized optics, they also plan to take advantage of pluggables anywhere they can because of the benefits they can provide in lower cost and power consumption.

Embedded Optics Used Primarily in Ultra Long-Haul Routes Operating Above 400G
On the performance optimized side, Glenn explained how it’s all about getting high performance, which by default usually means higher cost and proprietary implementations. He then went on to describe a field trial that Verizon completed where they were running embedded optics at maximum data rate for metro distances. They tested the same system at lower speeds for a long-haul route from New York to Chicago, demonstrating the benefits of a multi-haul solution. This proved that they could run 400G pretty much anywhere in the country.  Glenn also shared that on their long-haul networks, 80% is at or below 1,500km and 600G should be sufficient.

Verizon Plans to Continue Using Performance Optimized Embedded Optics for Ultra Long Haul
In the future, Glenn believes that both performance optimized optics and pluggables will continue to go up in performance. Today, the majority of Verizon’s new Ultra Long Haul (ULH) routes in its network are operating above 400G with embedded optics. According to Glenn, even if the next generation of optics gave Verizon 800G anywhere in the country with great performance, they would still want to know if they could get up to a Terabit or even 1.2T.  “If that will work on a lot of Verizon paths in a ULH environment, that is where Verizon is going to gravitate towards and it’s looking like that is going to happen with embedded optics,” added Glenn.

Acacia Sees Trends Influencing How Coherent Interfaces Should Be Developed
Tom Williams, Acacia’s Senior Director of Marketing, highlighted some of the key trends in the industry, starting with coherent moving to shorter reaches. This migration drives higher volumes as well as a need for more standardization because operators want the operational efficiency of interoperability and standards.

400ZR/ZR+ is exceeding expectations and is the fastest coherent product ramp ever displacing both proprietary coherent and direct detect solutions.

Tom recapped how 400ZR/ZR+ has exceeded expectations by being the fastest growing coherent ramp of all time. “We are seeing the core applications that drove the initial OIF 400ZR in the DCI use case and in many of those cases we saw MSA pluggables displacing performance optimized coherent and in others it replaced direct detect,” said Tom. “This has resulted in standardized pluggable optics taking up a higher portion of the coherent market.”

Tom also pointed out that during that same timeframe, the performance of embedded optics has scaled to the point where the industry is approaching the Shannon limit. As a result, component makers are now only seeing incremental improvements in spectral efficiency and operators are not getting significant improvement in capacity on the fiber. That has created a critical need to move to higher baud rates and higher data rates because they provide better economics and better power efficiency from optics. The most efficient way to do this is to develop high performance solutions that align with the investment that is already taking place in the shorter reaches in the pluggable space.

To illustrate the importance of taking the right steps with coherent development, Tom then described the path Acacia has taken. At each class, the company has developed both MSA pluggables and performance optimized solutions. Tom emphasized that, “The benefit that Acacia gets is that we make a big investment in each of these nodes, and we’ve been able to double the baud rate each time. We’ve seen the standards doing this and we’ve aligned our performance optimized solution with those standardized efforts.”  As an example, Acacia took the proven optics from the AC1200 operating at 70Gbaud and brought them into the company’s 400ZR/ZR+ products. Using the same platform allowed Acacia to then focus on advancing its packaging technology to take advantage of the efficiencies and have a scalable approach. And today, that same packaging technology is now feeding into the Coherent Interconnect Module (CIM) 8 module, which is enabling a pluggable performance optimized solution operating at 140Gbaud. It’s expected that the next generation will also have a doubling of the baud rates (as shown below) which would adhere to where the standards are expected to align.

Leveraging common development based on standards for both performance optimized, and pluggable solutions leads to higher data rates.

As Tom explained, the reason advancements in packaging have been so important is because the high-speed electrical interconnects are often the problem in the RF path when you start going to higher baud rates. By eliminating most of those interconnects, Acacia has been able to significantly extend the bandwidth of the RF chain and still get all the benefits of silicon and standards CMOS processes. Based on the Jannu DSP, the CIM 8 started shipping at the end of 2022 and has already been used in field trials with NYSERNet, China Mobile and Windstream Wholesale.

Nokia Looks to Lower Cost Per Bit with New Techniques
Serge Melle, Director Optical Product Marketing at Nokia, focused on what performance optimized solutions provide from a network applications standpoint. This started with a look at the trends that Nokia sees, and which are driving the key requirements for operators, which include bandwidth growth, faster internet speeds from IP routing platforms, and the need to lower network power consumption to achieve sustainability objectives.

As Serge stated, “This is really where this new generation of 5nm, 120Gbaud+ technology can address all of these things at the same time with the same technology evolution.” By being able to go to 1.2T wavelengths, operators can leverage the technology to achieve 2.4T in a single line card. That enables network operators to get much more scale in deployment of the bandwidth and reduce total cost of ownership, while also providing an effective means of transporting high speed services such as 400 Gigabit Ethernet and 800 Gigabit Ethernet. In addition, the performance improvement of this new generation of technology enables operators to triple the reach at 800G speeds – thereby going from several hundreds of kilometers to 2,000km distances at 800G per wavelengths.  Added Serge, “This is a very effective way to transport high-speed services across all network applications and can drive a significant reduction in the number of coherent optics needed in the network, which in turns leads to cost savings.  This technology can also provide significant network power savings because it can lower network power consumption by up to 60 percent.”

Like the prior speakers, Serge also stressed the importance of driving down the power per bit. “Every coherent generation has been able to achieve that and this latest generation of 1.2T technology continues the trend by leveraging 5nm silicon, silicon photonics and closer integration,” he said.

Finally, Serge shared how this latest generation of technology can be leveraged across a broad range of applications including metro DCI, long-haul and subsea networks.

Watch the replay
Click here to watch the replay. If you have any questions, please feel free to reach out to us via this contact form.