Bandwidth hunger continues to grow, which is why many enterprise data centers have set themselves the goal of migrating to 400G infrastructures as soon as possible.Large hyperscalers are already transitioning to 800GbE, a trend that will reach even large and medium-sized data centers faster than expected.Numerous vendors have already showcased their 800G support at the OFC 2022 show in San Diego.Users are faced with a plethora of standards and connector variants.Although the focus of San Diego's OFC 2022 (Optical Fiber Communication Conference and Exhibition) was calibrated on 400 Gigabit (GbE) Ethernet, 800GbE has already made its appearance.This arrival was not only demonstrated by the IEEE "Beyond 400G" session: exhibitors such as Accelink, Eoptolink, Infinera, InnoLight Technology, Linktel, Marvell, Source Photonics and Surinno Photonics arrived in California with 800 GbE transceivers, while MultiLane, Keysight Technologies , Spirent Communications and Viavi Solutions presented the corresponding test equipment.This fact is truly remarkable as 400 Gb ethernet only became commercially popular in 2019;it is evident that the demand for bandwidth is increasing at an ever faster rate.In 2019, network operators expected network traffic growth of 25% annually.Then came Covid-19 and consequently the home office instead of the office, video conferences instead of business trips, streaming TV instead of the cinema.Global data traffic has thus increased by 35% in 12 months.Market research firm LightCounting predicts that the five largest cloud and hyperscaler data center operators - Alibaba, Amazon, Facebook, Google and Microsoft - will see 200GbE and 400GbE transceivers peak in the next year and then start declining again. .According to this, in 2024 the sales of the first 5 customers with 800 GbE transceivers would exceed the sales of the previous two technology generations.But not only the big hyperscalers will need significantly more bandwidth in the future.Because the amount of data will increase dramatically in many other areas as well.Performance-hungry applications such as augmented and virtual reality (AR / VR), which in the coming years will be increasingly used in maintenance and service in industry, video inspections using drones, for example of wind turbines, tall towers and others inaccessible parts of buildings, quality control using cameras and many other moving image applications will generate significant amounts of data, especially if higher resolution video formats, such as in telemedicine, are also required.Campus 5G networks, which forward sensor and machine data to edge servers, and 5G public networks, which accelerate access to mobile data, also contribute to the growing data flow, as do AI-hungry applications. data that penetrate more and more areas of life.After the 800GBASE-R standard was approved by the Ethernet Technology Consortium in April 2020, it took less than a year to present the first suitable fifth generation DSP chips.In the meantime, a number of suppliers have presented the corresponding transceivers, such as Hitek Systems, Marvell or Microchip.The new generation of DSP, which is being produced for the first time in a 7nm CMOS process, is characterized by a number of innovations.With higher data rates (> 90 Gbaud), support for 800 G wavelengths and flexible adjustment of the required bandwidth, they enable more efficient network traffic.This allows network operators to reduce costs per bit per kilometer;the power requirement in watts per bit per km is also lower than in previous generations of transceivers.They can also configure more flexibly if they need more bandwidth or a longer range.The IEEE 802.3 Ethernet Working Group has established the P802.3df task force, which is to define the standards for 800 Gbps and 1.6 Tbps Ethernet.There will also be further standardizations for the 200 and 400 GbE variants.The further development of Ethernet technology allows for an increase in performance in two ways.Where previously 400 Gbps was possible via 8 lanes with 50 Gbps each (or gross: 56 Gbps), with 800 GbE you can use 8 lanes with 100 Gbps each (gross: 112 Gbps).Alternatively, the new standards make it possible to combine 2x400GbE into an 800GbE data stream.At the same time, the innovations also affect the previous Ethernet variants.With the new generation transceivers, 400GbE with 4x100 GBit / s will also be possible.And the next steps are already on the horizon.Next year they will realize 1.6TbE, then with 200 Gbit / s per lane (gross: 224 Gbit / s), which will also realize 800GbE with 4x200 Gbit / s.The International Photonics & Electronics Committee (IPEC) standardization body also presented basic specifications for the 800G DR8 and 2x400G FR4 interfaces at the end of January.The 800G DR and FR standards will follow in the second half of the year.IPEC's Physical Media Dependent (PMD) working group is working on a variety of network scenarios, including 800G broadcast over 500m, 2km, 10km and 80km.To take advantage of these bandwidth improvements, it was necessary to provide a simple transition from single-port to 400G dual-port optics in the 800G form factor, allowing the installed base of LC and MTP / MPO fiber connectors to continue to be used.Two competing industry consortia of transceiver manufacturers have submitted specifications to this effect, including connectors for 800GbE fiber cabling.OSFP MSA (Multi Source Agreement) added support for 800GbE via dual-400G and octal-100G to its OSFP module specification (current state: Rev 4.1) with Revision 4.0 in May 2021. However, the consortium points out that 800GbE was already possible with the previous specification.At the same time, work has begun on the future 1.6TbE, which will be introduced with Rev. 5.0.Almost at the same time, the QSFP-DD MSA group had extended its 6.0 hardware specification for the first time to include the QSFP-DD 800 standard, which is also based on 2x400G or 8x100G and will later be extended to 1.6TbE with 200 Gb per lane.This is to ensure a smooth transition from 400GbE to 800GbE and 1.6TbE.400 or 800G applications can be implemented in many different ways.This also increases the variability of the connectors.This trend was already evident with 400GbE and continues with 800GbE solutions.Because of this complexity, it is recommended that data center operators turn to cabling specialists such as Rosenberger OSI when implementing 400G or 800G applications and have a look at the specifications.After all, when choosing the right cabling solution, the application to manage is critical.With the 800G-SR8 multimode standard for short distances (up to 50m), the previous MPO / MTP technology will be used, optionally in the OM4 MPO / MTP 16 variant with beveled fiber cut (APC 8 °), or as dual MPO / MTP 4 + 4 (OCTO), with straight fiber cut (PC0 °).However, brand new Very Small Form Factor (VSFF) connectors such as MDC or SN are also on the way.In particular, US Conec's Miniature Duplex Connector (MDC), which can significantly increase port density and thereby save rack space, is believed to have great potential for the future.Developed as a Media Dependent Interface (MDI) or optical interface of the new SFP-DD and QSFP-DD transceivers, it may overtake the LC-Duplex connector in the future.For the singlemode standard 800G-DR8 (up to 2 or 10 km), corresponding MPO / MTP variants can also be envisaged, in the case of MPO / MTP 16 APC 8 ° or as double MPO / MTP 4 + 4 (OCTO) APC 8 ° .Not to forget the already mentioned MDC and SN connectors.For the 800G-DR4 variant for distances of 1 to 2 kilometers, the MDC or SN must also be used exclusively.The 800G-FR8 and 800G-LR8 standards can also be used to overcome distances of up to 2 or 10 km.Here manufacturers continue to rely on LC connectors, just like with the 800G-FR4, 800G-ZR and 800G-ZR-Lite.800G-2FR4 and 800G-2LR4 are offline, for which dual LC connectors are used, also in two different variants: either as single-mode duplex LC (LC-Compact) or as dual mini LC, also known as duplex mini LC.(Table 1)Development of the 800GbE is not finished yet: other standards have already been announced.The increase in bandwidth through higher transmission rates per link or new combinations during link grouping is also partly reflected in the standards for 200 GbE and 400 GbE.What this means for the infrastructure, which connectors are needed, and what the situation is with backward compatibility are questions that need to be answered.Save my name, email and website in this browser for future comments.