Gigalight


Font Size:big  middle  small

400G Technology on OFC 2018: Yes or No?

Number of visits: Date:2018-4-8 18:21

Abstract:

From 100G to 400G, it's inspiring for big data. But which company will seize the opportunity in the competition? That is a question. The advancement of 400G technology can win for more attention, more orders, more support for the companies. Therefore, nearly all optical communication vendors began the research and development of 400G products to capture the rapidly developing market.


The current situation is that a lot of companies demonstrated 400G products on OFC even though they know the 400G technology is immature. For reasons of secrecy, there is always no complete demonstration chain. Their explanation is that the demo is open only to key customers.

 

400g technology


We have reasons to believe that 56G PAM4 has indeed matured on the electric chip, but that doesn't mean the whole module is maturing. OFC2018 said that 100G PAM4 has also been maturing, which is impossible.


Big data needs millions of 100G optical transceivers per year, which let lots of optical module vendors pursue production rather than quality. The industry module test data from Gigalight within 4 months shows that even with 100G SR4, only two or three manufacturers have passed the reliability, and no one has achieved the reliability of >1000 hours for 100G CWDM4. People may think that reliability is not important, just as the 400G has not been developed successfully, it still can be used commercially. The whole optical communication industry is full of impetuous impulses.


400G, for its principle, is a combination of high - order modulation and DSP. 56G PAM4 has been demonstrated in the laboratory, but for 100G PAM4 high-order modulation, there will be two serious challenges, one is power consumption and delay, and the other is the received error rate. People can now modulate the three eye diagrams of the higher-order modulation, but they simply cannot receive data at the receiving end. In the future, the transmission of 100G PAM4 can be realized by using four levels. But if people want to import the technology to data centers for ultra-high density interconnection, it seems to be impossible.


In my opinion, the development of optical communication must be based on the various effects and potentials of the development of optics, and the various benefits and capabilities of electricity should be supplemented. If we take the opposite technical route, we will artificially introduce more complex algorithms and sacrifice the simplicity of the technology. In any case, the proposition is: which modality is better suited to the present needs and applications.


Cisco has indeed developed the switch with 32 400G ports, but the power consumption of each module is 13W. I believe this switch is only suitable for point-to-point or aggregation level networking, and it can't be used for data centers. Because the huge heat and electromagnetic interference, reliability issue and the necessity for heat accumulation and demagnetization will cause more delays for the entire system.


The mass transmission of information depends on density and speed. Density can use wavelength division multiplexing technology or multiplexing parallel technology. The challenge to speed involves modulation. If this modulation is to be complemented by DSP, the introduction of DSP will cause other cost problems, regardless of the availability of the technology. Therefore, it should be considered carefully for the use of PAM4 technology in data centers.


Just like other products, data centers also need to create low-cost applications and requirements. Even though we have money, we should not waste of money. Besides, data centers cannot pursue advanced technologies without destination. With PAM4 and Silicon Photonics technology, a grand blueprint for the future technology can be constructed. But there are too many advancing thoughts in OFC 2018 in order to cater to customers'  ideas and get more orders.


We must resolutely go on the long march of technological breakthroughs, but do not forget that the progress of human technology is one step at a time, and finally achieved the victory of the expedition. There is no such thing as a quick buck and a future built on a far less mature technology brick foundation, and finally a great success. This is not the logic of technology at all.

TypeInfo: Blog Articles

Keywords for the information:400G  OFC