400G Transceivers – Which form factor? (OSFP, QSFP-DD, CFP8)
The implementation of modern form-factors and features is not new with the introduction of new technologies. We are in the era of 400G, and like former cycles of new technologies, its market will bring in new transceiver form-factors with aims at specific applications of the network.
Next-generation form factors of transceivers have common attributes; undersized in form-factor, low power utilisation, and interoperability amongst all system dealers. The knowledge about the development of the 100G transceiver market will assist us in understanding the 400G technology launch.
Service providers needed a pluggable transceiver to facilitate extended reach and dedicated technologies like coherent detection. The data center team wanted a solution for a short range (about 2KM) application that entails not only low power consumption but also reduced cost.
The first 100G pluggable transceiver was the CFP form factor. Although it facilitated both short- reach and long-reach applications, it consumed a lot of space and power; 12Watts. However, as technology and components developed (concerning size and power consumption), the smaller CFP2 and CFP4 versions were launched into the market.
Even with improvements in technology, embedded lucid 100G and 200G technology are, hitherto, available only on CFP and CFP2 form factors.
Similarly, large-scale data centers, to meet their exceptional needs for more bandwidth capacity, have pushed the QSFP28 form factor for several short-reach applications (DAC, CWDM4, PSM4, and SR4). There has been more implementation of the QSFP28, and it provides a smaller form factor alongside lower consumption of power in comparison to the CFP form factors.
The Implementation of 400G Networks
The knowledge of the questions behind the applications of 100G will play a significant role in estimating the method of implementing 400G.
- Who needs the 400G pluggable transceivers?
- For which application?
- What about technology maturity?
- Is there any interoperability with former form factors?
With a similar logic as 100G, 400G is vital for the heavy-scale data centers and service provider, at the lower scale.
As 400G needs PAM4 modulation for transmission, the reach is even more challenging. The original transceiver reach of 400G will be restricted to only a few kilometers. For a more extended range, coherent detection, as well as its support technologies, is required alongside dispersion compensation, and amplification.
Similar to 100G, the 400G system will be adopted based on the intended applications; it is most likely a dedicated form-factor for the data centers and another one for longer-reach applications. The previous 400G technology development avoids the transitional form-factors (CFP2, CFP4), which the adoption of 100G followed.
The 400G launch will occur alongside two form-factors for the access networks and data centers.
- QSFP56-DD (also known as QSFP-DD for QSFP Double Density)
- OSFP (for Octal SFP)
Equally, the two form factors are running 8x 50G PAM4 lanes on the electrical side, while it can either be eight 50G PAM4 lasers or four 100G PAM4 lasers on the optical hand.
- OSFP facilitates higher power (< 15W*) compared to the QSFP-DD (<12W*). The OSFP supports an early implementation since it is less challenging to release a 15W-designed technology than 12W-designed version.
- While the QSFP-DD port is backward-compatible with QSFP+ (40G), QSFP28 (100G) and QSFP56 (200G), the OSFP port uses a QSFP to OSFP converter module.
- While the OSFP integrates thermal management unswervingly into the form-factor, the QSFP-DD doesn’t.
Equally, QSFP-DD and OSFP are intended for intra-DC applications together with DAC, AOC and optical connection (up to 2km). Additional options are being developed to support data center interconnect (DCI) with extended reach, and other technologies such as DWDM super channel.
The CFP8 form-factor, a definition of the CFP MSA, is fundamentally dissimilar than QSFP-DD and OSFP:
- Facilitating up to 24W power consumption
- It possesses 16x channels 25G NRZ on the electrical side (in place of 8x 50G PAM4 intended for QSFP-DD and OSFP)
- Supplying for an MDIO management interface (rather than I2C meant for QSFP-DD and OSFP)
The CFP8 is meant for transmission application, with its wide-spread footprint and high consumption of power (up to 24W). The first version (CFP8 400GBASE-LR8) will support about 10Km, implementing 16x 25G NRZ electrical lanes that are converted to 8x 50G PAM4 paths.
Further alternatives that use the technology of coherent detection, which will support a distance of around 80 Km, are coming up. Apart from facilitating longer distances, the CFP8 is the foundation of 800G. It is possible to have 800G by combining to the 16x electrical lanes of CFP8, 50Gbps PAM4 modulation DSP, multiplexing of lasers and coherent detection. It is unarguable that the future for this technology is still far off.
We are in the era of 400G. The knowledge of technology applications will help heighten our understanding of the adoption of transceiver technology.