Huawei CloudEngine Switches

2019: Switch of DPC Huawei CloudEngine 16800

On January 14, 2019 the Huawei company provided the switch of data processing centers created for an era of the artificial intelligence (AI) – CloudEngine 16800. Huawei considers AI defining for an era three characteristics of switches for DPC: the built-in AI processor, network interface card 400GE with 48 ports and capability to evolve towards off-line control. The company includes all these technologies in the switches of data processing centers, helping clients to accelerate digital transformation from business.

According to the report of Huawei Global Industry Vision (GIV) 2025, the coefficient of implementation of artificial intelligence technologies will grow by 2025 to 86% whereas in 2015 it made 16%. Capability it is effective to use artificial intelligence for change of business models, decision-making and quality improvements of service will become, according to Huawei, key driving force.

"Before emergence of the perfect intellectual world there was not enough time. Data processing centers become a core of such infrastructures as a 5G and AI. Implementation of artificial intelligence technologies in switches of data processing centers, promotes transition of DPC networks of an era of cloud computing to an era of artificial intelligence". Kevin Hu, president of the Network direction of Huawei company

With approach of an era of artificial intelligence, its computing power becomes dependent on performance of DPC networks that becomes a key problem of commercial implementation of AI technology. Within traditional Ethernet technology, the computing power of AI in data processing centers reaches only 50%, noted in Huawei. At the same time in the company assume that annual amount of data worldwide will grow about the 10th zettabyte in 2018 to the 180th zettabyte (180 billion terabyte) in 2025. According to representatives of Huawei company, the existing DPC networks of standard 100GE will not cope with such data stream. Besides, traditional methods of non-automated maintenance will not meet the increasing requirements as the number of servers of DPC continues to grow, and in parallel there is consolidation of networks. Respectively, it is obviously necessary to develop and implement technologies for performance improvement of intelligent systems of maintenance.

According to the statement of the producer, CloudEngine 16800 there is a switch of a data processing center with the built-in high-performance AI processor using the algorithm iLossless providing automatic optimization under traffic patterns that provides a low delay and increases capacity. The CloudEngine 16800 switch overcomes the restrictions of computing power caused by loss of packets in traditional Ethernet thereby increasing the computing power of artificial intelligence from 50 to 100 percent and increasing the number of input-output operations per second (IOPS) of the data warehouse by 30 percent.

Network interface card 400GE with 48 ports shows the high density of ports and will help to cope with fivefold increase in traffic during an era of artificial intelligence, the producer claims.

As reported in Huawei, CloudEngine 16800 is created on an advanced hardware switching platform – thanks to orthogonal architecture it avoids the traditional technical issues connected with high-speed transmission of a signal, heat emission and power supply. Besides, energy consumption on 1 bits decreases by 50% that ensures more eco-friendly functioning. Thanks to the local system of the intellectual analysis CloudEngine 16800 and the centralized network FabricInsight analyzer, the distributed AI system of maintenance reveals failures in read seconds and automatically determines their sources, thereby accelerating autonomism of network. Besides, this architecture increases flexibility of the systems of maintenance.

2018: Description of a line of CloudEngine

Infrastructure specialists and operation of networks of the increasing number of the enterprises implement hardware SDN platforms for quality improvement and efficiency of client service. It is supposed that similar platforms should satisfy requirements of the enterprises at different stages of transition from traditional data centers to private and hybrid clouds. It means that hardware platforms should be open, flexible and capable to be upgraded for the next 5-7 years.

For March, 2018 switches of a line Huawei CloudEngine provide extensive product assortment and also create the unified, open and programmable architecture providing trouble-free transition from traditional networks of data centers to SDN and a multicloud environment. Besides, they provide the standard API interfaces for interaction with third-party computing platforms, controllers and instruments of automated management for the purpose of satisfaction of various customer needs and creation of opportunities for the integrated implementation.

From the moment of emergence in the market in 2012 the solution Huawei CloudFabric and switches of a line CloudEngine show the potential of growth and are actively applied in commercial practice in combination with the advanced network platforms in such segments as providing cloud services, finance, mass media and entertainments and telecommunications. For March, 2018 the solutions Huawei CloudFabric are implemented by more than 2,800 enterprises in 120 countries of the world.