To cope with large-scale cloud services and other growing data center storage and processing needs, the data center systems have become increasingly decentralized and difficult to manage. And applications like artificial intelligence (AI) urgently need low-latency, high-bandwidth network architectures to support the large number of machine-to-machine input/output (I/O) generated between servers. To ensure the basic performance of these applications, the maximum fiber propagation between these distributed data centers must be limited to about 100 km. Therefore, these data centers must be connected in distributed clusters. In order to ensure high-bandwidth and high-density data center interconnection at the same time, 400G ZR came into being. In this post, we will reveal what 400ZR is, how it works and the influences it brings about.
400ZR, or 400G ZR, is a standard that will enable the transmission of multiple 400GE payloads over Data Center Interconnect (DCI) links up to 80 km using dense wavelength division multiplexing (DWDM) and higher-order modulation. It aims to ensure an affordable and long-term implementation based on single-carrier 400G using dual-polarization 16 QAM (16-state quadrature amplitude modulation) at approximately 60 gigabaud (Gbaud). Developed by Optical Interconnect Forum (OIF), the 400ZR project is essential to facilitate the reduction of the cost and complexity of high-bandwidth data center interconnects and to promote interoperability among optical module manufacturers.
Figure 1: 400G ZR Transceiver in DCI Switch or Router
400G ZR proposes a technology-driven solution for high-capacity data transmission, which could be matched with the 400GE switch port. It uses a unique design of advanced coherent optical technology for small, pluggable form factor modules. Although the product form factor is not specified in the IA (implementation agreement), the companies or groups contributing to the 400ZR have defined this specification to fit the solution. These form factors defined separately by Multi-Source Agreement (MSA) bodies specify compact mechanical transceivers like QSFP-DD and OSFP, which are connectorized and pluggable into a compatible socket in a system platform. That is to say, the compatible 400ZR solutions that come to market will also be interoperable since the OIF and form factor MSAs are industry-wide organizations. And the interoperability of the 400ZR solutions offers the dual benefit of simplified supply chain management and deployment.
Like other 400G transceivers, the pluggable coherent 400ZR solution can support 400G Ethernet interconnection and multi-vendor interoperability. However, it is not suitable for next-generation metro-regional networks that need transmission over 80 km with a line capacity of 400 Gb/s. Under such circumstances, 400ZR+, or 400G ZR+ is proposed. The 400ZR+ is expected to further enhance modularity by supporting multiple different channel capacities based on coverage requirements and compatibility with installed metro optical infrastructure. With 400ZR+, both the transmission distance and line capacity could be assured.
Although 400ZR technology is still in its infancy, once it is rolled out, it will have a significant impact on many industries as the following three: hyper-scale data centers, distributed campuses & metropolitan areas and telecommunications providers.
The development of DCI and 400ZR could help cloud and hyper-scale data centers adapt to the growing demand for higher bandwidth on the network. They could deal with the exponential growth of applications such as cloud services, IoT devices, and streaming video. As time goes by, 400G ZR will contribute more to the ever-growing applications and users for the whole networking.
As is mentioned above, 400ZR technology will support the necessary high-bandwidth interconnects to connect distributed data centers. With this connection, distributed data centers can communicate with each other, share data, balance workloads, provide backup, and expand data center capacity when needed.
400G ZR standard will allow telecommunications companies to backhaul residential traffic. When running at 200 Gb/s using 64 Gbaud signalings and QPSK modulation, 400ZR can increase the range of high loss spans. For 5G networks, 400G ZR provides mobile backhaul by aggregating multiple 25 Gb/s streams. 400ZR helps promote emerging 5G applications and markets.
In addition to the interoperable 400G mode, the 400ZR transceiver is also expected to support other modes to increase the range of addressable applications. These modes are called 400ZR + and 400ZR-. “+” indicates that the power consumption of the module exceeds the 15W required by IA and some pluggable devices, enabling the module to use more powerful signal processing technology to transmit over distances of hundreds of kilometers. “-” indicates that the module supports low-speed modes, such as 300G, 200G, and 100G, which provide network operators with more flexibility.
According to the data source below from LightCounting, 400ZR will lead the growth of optical module sales in 2021-2024. The figure below shows the shipment data of high-speed (100G and above) and low-speed (10G and below) DWDM modules sold on the market. It is clear that modules used in Cloud or DCI have an increasing tendency in 2021-2024. That means 400ZR will lead annual growth from 2021.
In addition, with the first 100Gbps SerDes implementation in switching chips expected in 2021, the necessary data rate will move to 800 Gbps within the next 1-2 years for the optics interface. Since the OSFP form factor has been defined to allow an 8x 100GE interface without changing the definition of the transceiver. Similarly, in parallel, the coherent optics on the line side will transition to support 128GBaud 16QAM within a similar time frame, making it easy to migrate from the current 400ZR to the next-generation 800ZR. Therefore, 400ZR is crucial no matter in the current or the future network development.