• Technical Conference:  30 March – 03 April 2025
  • Exhibition: 01 – 03 April 2025
  • Moscone Center, San Francisco, California, USA

Cloud Data Center Evolution - From 0 to 400G

By Macom

Cloud services and consumer products that rely on Cloud Data Centers have created a rapid evolution in Data Center capabilities. As services like the Internet of Things (IoT), audio/video content delivery, big data processing and social media continue to demand greater speeds, the Cloud Data Center industry is making strides to create solutions that are energy efficient and easily expandable. As a result, the need for low cost 100G is creating a shift in how the industry approaches the backbone hardware that enables Cloud Data Centers to respond and expand for both commercial and consumer needs. As Cloud Data Center providers work to find energy efficient low-cost solutions, the industry is looking toward future requirements that meet these needs as well. The evolution from 0 to 100G is just the beginning.

The Need for Low Cost 100G
The above stated increased traffic within and between Cloud Data Centers is driving the need for low cost 100G, moving to high-speed and low power 200G and 400G interconnects. Starting in 2018, Cloud Data Center OEMs will adopt new technology, allowing them to increase bandwidth density per port. Smaller QSFP, QSFP-DD and OSFP form factor modules will support these interconnects. The shift requires suppliers to deliver lower power electronic components that are reliable over time. One solution is to drive chipset innovations that address the power and bandwidth density needs by supporting single wavelength interconnects. When combined with silicon photonics, this approach fulfills the power envelope requirements of these smaller form factor modules.

New Capacity Requirements
Looking back, transceiver technology was expensive and power hungry. Silicon technology advances over time allowed for more affordable hardware and energy-efficient operations. But the limits to the capabilities of the current transceiver architecture are already here and today's Cloud Data Centers are at maximum capacity.

On average, large scale Cloud Data Centers need to upgrade networking hardware every two years. With the switches that are currently in use, the cost of the optical transceivers is a major contributor to the upgrade cost. Next generation 100G solutions using PAM-4 technology can help address these issues and lower the cost per bit in the future.

This new approach to 100G has benefits in the near and long-term. It increases density, lowers power and lowers cost per transmitted bit, thereby significantly improving overall efficiency. As companies continue to grow in scale and their data needs become more complex, 100G per lambda will be a building block to next generation 400G, which offers the bandwidth and efficiency they desperately need.

100G PAM-4: The best solution for 100G
Understanding the current market challenges and the growing need for speed shows that PAM-4 is the best approach for a scalable solution. 53Gbaud PAM-4 modulation using mixed signal PHYs can address the challenge from the industry and provide lower cost hardware and more energy efficient solutions over time. For 100G transceivers, single-wavelength PAM-4 technology reduces the number of lasers to one and eliminates the need for optical multiplexing. The 2 bits / symbol approach of PAM-4 reduces the bandwidth requirements, while improving cost and power per bit. Digital Signal Processing (DSP) provides flexible and adaptable equalization, allowing 100G transmission over single mode fiber up to the distance of 2 Km.

PAM-4 proves to be the most cost-effective, efficient enabler of 100G and 400G in the Cloud Data Center to date. For 400G implementations,only four optical assemblies are needed, representing a major opportunity for Cloud Data Center operators to reduce their CAPEX and OPEX with an extremely compact and energy efficient module.

The Best Solution
Research and testing shows that single lambda PAM-4 supports the increased speed requirements of Cloud Data Centers and is achievable with technology that exists today. PAM-4 and the shift in approach to 100G provides a 60% reduction in component count along with a 33% reduction in power requirements. The significant reduction in assembly costs and higher reliability deliver initial and long-term value and provide the infrastructure required to reach 400G as well. At 400G, this technology is adaptable and can enable QSFP-DD or OSFP form factor transceivers. The PAM-4 approach delivers the best overall solution for speed and affordability in Cloud Data Centers.

Optical and Electrical Results
With its streamlined architecture, lower cost and higher reliability, single lambda 100GE PAM-4 or 100G Serial, can be seen as the equivalent of SFI at 10GE. SFI and SFP+ enable the cost reduction and high density required to drive the growth of 10GE.

The single lambda 100G solution offers the optimal component count with the simplest architecture. This brings the potential to achieve the lowest cost as a result. In addition, reducing the optical components to the minimum required set enhances module reliability, manufacturing yields and reduces the chance of optical component failure. 100G per lambda is poised to create the next wave of explosive growth in mega Cloud Data Centers. 

*Direct from the source, Macom

Posted: 16 November 2017 by Macom | with 0 comments

Blog post currently doesn't have any comments.
 Security code

The views expressed in this blog are those of the authors and do not necessarily reflect the views or policies of The Optical Fiber Communication Conference and Exposition (OFC)  or its sponsors.