400G Ethernet Transceivers – Introduction

400G Ethernet Transceivers introduction is so important right now, as 400G Ethernet is the big thing that everybody is and will be talking about for years to come. 400 GbE is the next big step since 100GE. Being and staying informed and up to date with the 400G era interfaces is a must. That is why the next paragraphs will go over:

1. 400G Ethernet Transceivers Demand, Driving Force, and Standard
2. DSP Empowered 400G
3. PAM4 Basics
4. 400G Ethernet Transceivers Capable Form-Factors (QSFP-DD, OSFP, CFP8)
5. 400G QSFP-DD Transceiver Roots in 100G QSFP28 Modules
6. 400G QSFP-DD Transceiver Types
7. Road to the Next Big Leap
8. Summary

1. 400G Ethernet Transceivers Demand, Driving Force, and Standard

Since as early as the beginning of 2018, Demand for 400G Optics has been present in Tier 1 Data Centers. The main driver for this has been Data Centers that host and utilize cloud-based services, and an increase in the number of Internet-connected devices and users. It is not a secret that IoT (Internet of Things) is and will increase dramatically, with devices becoming more reliant on exchanging data with other devices and systems over the Internet. To improve the throughput capabilities and reduce latency, 5G NR (New Radio) technology has been developed to ensure the performance of Fronthaul and Backhaul. More on this in Metro Transport Network in 2019 and CPRI vs OBSAI articles.

Due to the massive intra- and inter-data center traffic, IEEE 802.3bs, defining 400GbE over optical physical media, was introduced in late 2017. With a more recent addition of IEEE 802.3cm defining 400GbE over multimode fiber, utilizing 4 or 8 transmission lines. However, several additional standards are underway to be approved in 2021.

2. DSP Empowered 400G

Previously, to upgrade from a 100G to a 400G connection, you needed to add extra communication lines performing at 100G. Now it is possible to transition from 100G to 400G by simply upgrading each side of the connection without touching or adding new lines of communication. This is possible due to Monumental improvements in DSP (digital signal processor) quality and production costs. DSP acts as a glue to connect the transceiver's electrical and optical interfaces. It is a Gearbox that converts, for example, 8 electrical channels of 50G PAM4 signals to 4 optical channels of 100G PAM4 signals. Nevertheless, DSP can utilize PAM4 due to receiving a CRC retimer and FEC Block integration to make sense of all the noise.

3. PAM4 Basics

From this, a question can come up – What is PAM4? In its essence, PAM4 is what it says, Pulse Amplitude Modulation 4-level. It allows for to transfer of two bits at the same time.

This is a significant improvement over the most common encoding for 5G enabling transceivers like 25G SFP28 and 100G QSFP28 that use NRZ (Non-Return-to-Zero), i.e., PAM2 (Pulse Amplitude Modulation 2-level).

However, there are important drawbacks to PAM4 when compared to NRZ. PAM4 increases SNR (Signal-Noise Ratio) by -9.54 dB (Link Budget Penalty), and higher requirements for link quality are mandatory. For example, clock stability and receiver quality. More on this in the 100G Single Lambda [https://edgeoptic.com/100g-pam4-single-lambda/] article.

4. 400G Ethernet Transceivers Capable Form-Factors (QSFP-DD, OSFP, CFP8)

When talking about transceivers, form-factor and its capabilities play a vital role. Adding to this, widespread adoption of 400G Ethernet Transceivers is the next big step. That is why there is a debate on the best-suited factor for 400G. The main competing form factors are QSFP-DD, OSFP, and CFP8. 400G Transceiver Form Factors shows a comparison between said three form factors.

Each form factor has advantages and drawbacks. There will be applications and solutions that will utilize one of these QSFP-DD, OSFP, or CFP8 form factors to ensure maximum efficiency.

However, QSFP-DD at the current state looks like a favorite due to how much of the industry is in love with the QSFP form-factor and 100G QSFP28 modules. Additionally, QSFP-DD provides the greatest Port Density compared to the competition. This gives QSFP-DD manufacturers much greater appeal than the new OSFP form-factor modules and bulkier CFP8 modules. Furthermore, QSFP-DD is developed to ensure backward compatibility with QSFP from-factor ports. Allowing for a more gradual update from upgrade 100G QSFP28 systems to 400G capable solutions.

Nevertheless, Power Consumption – Maximum amount of Power that can be dissipated by the module without damaging internal components – is of great importance. Main Power Consumption comes from the Micro-Controller and PAM4 DSP that have to perform at mind-boggling speed. Solutions are developed to improve the efficiency of QSFP-DD to ensure internal component heating will be reduced below the critical point. An upside is the emergence of practical and applied EUV (Extreme Ultraviolet Lithography) that will reduce the power draw of logic components.

5. 400G QSFP-DD Transceiver Roots in 100G QSFP28 Modules

All of this leads to understanding the 400G QSFP-DD (Double Density) form-factor and technology used to ensure connection between both ends of the line. As previously mentioned in the DSP section, 400G is the upgrade from 100G that is enabled by improved DSP and the use of PAM4. However, it is also different from QSFP28 by having 4 additional Tx and Rx lanes that require an additional of one extra electrical connector to ensure the space for said Tx and Rx lanes. This results in 8 lanes of 50G PAM4 traffic between the host and the pluggable 400G QSFP-DD. From there, DSP takes over to produce the required optical 4 lanes of 100G PAM4 optical signal. (400G QSFP-DD SR8 converts 8 electrical signal lanes to optical lanes, both using 50G PAM4.)

6. 400G QSFP-DD Transceiver Types

The QSFP-DD and QSFP28 similarities do not stop there, as technology is also adapted. Below 400G Transceiver Types shows the main 400G QSFP-DD parameters and is sorted by Industry Name.

The easiest way to see differences in technology that is applied is by looking at the Wavelength column. It can be split into five parts – SR8, PSM4, CWDM4, LWDM4, and LWDM8. Except for SR8 and LWDM8, which add to lanes or wavelengths used, all of the technology principles are brought over from QSFP28. More on these principles can be read in the 100G QSFP28 module types and main differences article. Additionally, MPO [Split channel design] is used for shorter distances, and LC [Single multiplexed channel] design is used for longer distances, which currently are up to 40km with 400G QSFP-DD 80km modules around the corner.

7. Road to the Next Big Leap

Although 400G QSFP-DD 80km sounds impressive, the future is even grander in its scale. Adding to this is the increase in working or studying remotely or from home. This has lit a huge fire under the pot. This can be seen in Cisco Projections 2017, which expected a 26% year-over-year growth for IP Traffic.

However, this was broken in the year 2020. In just one month, traffic increased by a year’s worth, as seen in the Cisco Projections 2017.

As mentioned before, the pot has been smoking to keep up with this staggering change. Only going faster, upgrading from 100G to 400G, may not be enough to ensure that everyone stays connected. Thus, a call for 800G, 1T, 1600G has been heard, and the industry is under a timer to increase the bandwidth once more.

8. Summary

In short, 400G has proven the ingenuity of mankind once more. It has been a long journey since the inception of the First Ethernet experiments by Bob Metcalfe in 1972. Now 1G, 10G, 40G, and 100G sound self-evident and of no importance. However, all of these achievements have served as stepping stones for The World of Today with bright hopes for the future. We highly appreciate the time you spent reading this, and if you have any questions or comments, feel free to reach us by email sales@edgeoptic.com or by giving us a call at +371 22084457.