RADHA NAGARAJAN 400G DCI
ENABLING A SEAMLESS TRANSITION TO 400GDCI INFRASTRUCTURE
As the industry plans its migration of DCI infrastructure to 400G pluggable technology, the common view has been that different solutions will be used at the edge and in metro and long-haul networks, writes Radha Nagarajan , CTO, Interconnect, Inphi.
T he emerging concept is to use the same 100G/200G/400G is nearing reality with the availability of low-power, high-density QSFP- DD, OSFP and CFP2-DCO coherent pluggable modules that support the Optical Internetworking Forum (OIF) 400ZR standard. These modules will drive a paradigm shift in how DCI is deployed by cloud and telecom service providers, enabling high-density IP over DWDM on switch and router platforms that also have the performance required for metro and long-haul networks. Motivated by the need to lower cost and power consumption of DCI, leading cloud operators, OEMs, module and chip vendors have joined the OIF in the development of an industry standard for 400ZR compact pluggable modules. There has been an evolution in thinking as the industry now contemplates where 400 ZR fits in a “DCI- pluggable digital coherent optics (DCO) module across all distances. This concept
Solving each of these challenges required the development of an industrywide ecosystem that supports pluggability and a new era of interoperability. The ecosystem that has evolved will enable telecom and cloud providers to not only deploy solutions for DCI edge applications up to 120km but also use the same 100/200/300/400G solutions across the entire DCI infrastructure. Several new developments are paving the way for pluggable modules, including transceiver integration and interoperability advances that enable this seamless transition to 400G. These advances are being driven in several ways. The first is through the availability of 400G single-chip, coherent Silicon Photonics Integrated Circuit (PIC) devices that include all transmit and receive functions, and low-cost methods for passive alignment of fibre to this PIC device that eliminates the complicated active alignment, using traditional optics. Another important development has been the integration of an industry-standard firmware management interface in transceivers that enables full performance monitoring. This was previously only available in dedicated transport systems, directly from the optical module. Equally significant is the availability of low- power, high-performance, 7nm CMOS- based coherent DSPs that enable 400ZR as well as extended reach 100/200/300/400G ZR+ modes. The use of probabilistic shaping technology in these latest DSP architectures enables them to maximise data rate at longer fibre distances while delivering lower deployment cost per bit. The DSPs also leverage new forward error correction (FEC) PAVING THEWAY
Any-Distance” approach (Figure 1). There also have been some obstacles to navigate that, until now, have made it difficult to provide the seamless transition to coherent 400G pluggable solutions for routers and switches. First, coherent DSP ASICs have, to date, been designed for dedicated transport boxes. The power consumption and package size had not been sufficiently optimised for QSFP- DD or OSFP form factors preferred in the high-density switch and router interfaces. These DCI dedicated boxes add cost and dissipate power. Second, there has been a lack of Interoperability across generations of coherent solutions for 100G and 200G. With the switch and router ports evolving to 400G, there was a need for an industry wide 400G coherent standard that was interoperable. And third, there has been a lack of a clear application space, as the conventional coherent technology had been deployed in many forms across various distances and optical fibre span configurations.
Figure 1: Technology choices in optical transport for datacentre interconnects.
14
| ISSUE 20 | Q1 2020
www.opticalconnectionsnews.com
Made with FlippingBook Ebook Creator