Will GPON and EPON get along at 25G? Only if China Telecom rules
Julie Kunstler, Principal Analyst, Components and Next-Gen Infrastructure
China Telecom is the only major FTTP operator with both EPON and GPON equipment deployed in its network. Most operators would try to avoid this situation, but GPON was not ready for mass deployment when China Telecom began to implement FTTP. Skip ahead several years and China Telecom has begun to upgrade its 1G EPON network with 10G EPON and will begin to upgrade its 2.5G GPON network with 10G GPON. The divergent standards continue to consume resources at China Telecom because product and interoperability testing is needed for both next-gen EPON and GPON. Several equipment vendors are trying to converge EPON and GPON standards at 25G. However, this is not the first time that several vendors have attempted to align the two standards bodies; efforts at 10G failed. Ovum believes that China Telecom is tired of spending resources on parallel testing and evaluation. This frustration came through during China Telecom's presentation for the "Lessons Learned From Global PON Deployment." Given China Telecom's dominant position in the consumption of PON components and equipment, it has the power to force vendors to supply converged solutions and thereby bypass the standards bodies. While this approach would require China Telecom to spend resources on 25G PON specifications, it would provide a single upgrade path, thereby saving resources in the long term.
IP-over-WDM rises again
Don Frey, Principal Analyst, Transport and Routing
With 10G and 40G, some carriers deployed IP-over-WDM to lower capex, but suffered from integration and support of alien wavelengths (optical signals originating from equipment not under their direct control). In the 100G world, adoption of IP-over-WDM has been slow, as heat and power requirements created additional challenges. Now, with improved heat and power performance from new, single-wavelength coherent solutions and SDN control, optical integration into switches and routers is coming back into fashion. Juniper announced a 1.2Tb WDM card capable of six 16QAM/8QAM signals for its QFX data center switch product line, currently the densest per slot integrated WDM-plus-IP product in the market. There is still a performance penalty because the WDM card does not fully support the total switching capacity, but it does remove the need for an expensive, additional optical blade. The target application for this solution is the DCI (data center interconnect), as switch power and heat usage in the data center (DC) is a bit more forgiving than in the telecom space. Juniper's solution now takes its place alongside offerings from Arista (600G) and Facebook Voyager (800G). All three have coherent optics–based DC switches. DC operators will need to evaluate these solutions to determine the best fit.
Nokia and Facebook show beyond-100G prowess across the Atlantic
Ian Redpath, Practice Leader, Components, Transport and Routing
Nokia and Facebook successfully completed multiple submarine field trials over a 5,500km submarine cable between New York and Ireland. This submarine field trial of Nokia Bell Labs' new probabilistic constellation shaping (PCS) technology used shaped 64QAM, which achieved a record spectral efficiency of 7.46b/s/Hz and increased the stated capacity of the system by almost 2.5 times. The test also included a successful 11,000km roundtrip submarine transmission using shaped 64QAM, achieving a record spectral efficiency of 5.68b/s/Hz and a first demonstration of 200Gbps and 250Gbps wavelengths and 16QAM modulation over a transatlantic submarine route using a real-time coherent DSP.
The Nokia–Facebook trials are significant from a number of perspectives. First, by working with Facebook, Nokia has placed itself in a strong position to profit from the fast-growing, internet content provider–fueled wave of subsea cable investments. Nokia has demonstrated the versatility of its PSE-2 chipset and broadened the beyond-100G application space to include subsea. Nokia has also shown technological strength with 8QAM, 16QAM, and 64QAM transmission at transatlantic distances.
Squeezing long-haul optics into an OSFP module: A Google wish
Kevin Lefebvre, Principal Analyst, Components, Transport and Routing
During the "Hyperscale Data Centers: Inside and Out" panel discussion, Ryohei Urata from Google was asked whether OSFP (octal small form factor pluggable) or QSFP-DD (quad small form factor pluggable double density) was going to lead in the intra–data center market. Ryohei outlined his vision of putting long-haul coherent optics in a client-side pluggable for speeds beyond 400G. This would allow Google to use the same module for inter- and intra-data center distances while also giving it the ability to use coherent optics within the data center. Google would expect the switch manufacturers to upgrade to the new form factor, which would eliminate the need for backward-compatibility with QSFP28. Google strongly favors the OSFP form factor. However, Ovum believes that placing coherent optics and a DSP within an OSFP module would be a huge challenge and require substantial R&D investments. Managing the power and heat within the module alone would be extremely difficult. The industry has been trying for years, unsuccessfully, to place long-haul optics into a client-size pluggable. To help the industry move forward, Google might need to assist with NRE (non-recurring engineering) funding.