Chinese telecommunication vendor Huawei has unveiled the industry's first data center switch built for the Artificial Intelligence (AI) era, which has been named as the ‘CloudEngine 16800’, at its network product launch event for Spring 2019 themed "A CloudEngine Built for the AI Era".
Huawei defines three characteristics of data center switches in the AI era: embedded AI chip, 48-port 400GE line card per slot, and the capability to evolve to the autonomous driving network, and innovatively incorporates AI technologies into data center switches. The pervasive use of AI will help customers accelerate intelligent transformation.
According to Huawei's Global Industry Vision (GIV) 2025, the AI adoption rate will increase from 16% in 2015 to 86% in 2025. The capability of leveraging AI is to reshape business models, make decisions, and improve customer experiences will become a key driving force. Kevin Hu, President of Huawei Network Product Line, said, "A fully connected, intelligent world is fast approaching.
Data centers become the core of the new infrastructures such as 5G and AI. Huawei will first introduce AI technology to data center switches, leading data center networks from the cloud era to the AI era."
With the advent of the AI era, the AI computing power is affected by the performance of data center networks, which is becoming a key bottleneck of the AI commercial process.
On a traditional Ethernet, the AI computing power of data centers can only reach up to 50 percent due to a packet loss rate of 1/1000. At the same time, the industry expects that the annual volume of data worldwide will increase from 10 zettabytes in 2018 to 180 zettabytes (180 billion terabytes) in 2025. Existing 100GE data center networks will be unable to handle the predicted data flood.
In addition, traditional manual O&M methods will be unable to meet requirements as the number of data center servers continues to increase and the computing network, storage network, and data network become converged. It is therefore imperative that innovative technologies be developed and introduced to improve the intelligent O&M capability.
"The data center switch built for the AI era has three characteristics," said Kevin Hu, namely "embedded AI chip, 48-port 400GE line card per slot, and the capability to evolve to the autonomous driving network." Industry's first data center switch with an embedded AI chip, reaching an AI computing power of 100%.
The CloudEngine 16800, the first data center switch in the industry to leverage the power of an embedded high-performance AI chip, uses the innovative iLossless algorithm to implement auto-sensing and auto-optimization of the traffic model, thereby realizing lower latency and higher throughput based on zero packet loss.
The CloudEngine 16800 overcomes the computing power limitations caused by packet loss on the traditional Ethernet, increasing the AI computing power from 50 percent to 100 percent and improving the data storage Input/Output Operations Per-Second (IOPS) by 30 percent. Industry's highest density 48-port 400GE line card per slot, meeting the requirements for 5-fold traffic growth in the AI era
The CloudEngine 16800 boasts an upgraded hardware switching platform, and with its orthogonal architecture, overcomes multiple technical challenges such as high-speed signal transmission, heat dissipation, and power supply.
It provides the industry's highest density 48-port 400GE line card per slot and the industry's largest 768-port 400GE switching capacity (five times the industry average), meeting the traffic multiplication requirements in the AI era. In addition, the power consumption per bit is reduced by 50%, ensuring greener operation.
Enabling the autonomous driving network, identifying faults in seconds, and automatically locating faults in minutes
The CloudEngine 16800 is embedded with an AI chip, greatly enhancing the intelligence level of devices deployed at the network edge and enabling the switch to implement local inference and rapid decision-making in real time. With CloudEngine 16800's local intelligence and the centralized network analyzer FabricInsight, the distributed AI O&M architecture identifies faults in seconds and automatically locates the faults in minutes, helping to accelerate the advent of autonomous driving network. Furthermore, this architecture greatly improves the flexibility and deployability of O&M systems.
Leon Wang, General Manager of Huawei Data Center Network Domain, said, "Huawei CloudEngine series data center switches have been successfully launched into commercial use for more than 6000 customers, helping digital transformation of industry customers such as finance, Internet, and carrier customers. Huawei launched the CloudEngine 16800 to help customers accelerate intelligent transformation, achieve pervasive use of AI, and jointly build a fully connected and intelligent world."