Easily release potential and popularize cloud native and artificial intelligence through Qingyun Cloud

Feixiang News (Wei Deling/Text) Cloud native and AI are the two words most likely to be mentioned in the technology circle in 2023. In the Chinese ICT technology released by Gartner not long ago In the maturity curve, these two technologies are also at the peak of the curve, in the stage of inflation of expectations, which indirectly proves the current popularity of these two technologies and their immeasurable future.

In addition, in many cases, these two words are often mentioned simultaneously. Cloud native provides application flexibility and rapid deployment, providing opportunities for artificial intelligence to be quickly implemented on different infrastructures. For example, in traditional communication systems, cloud-native containerization provides traditional devices with a wealth of optional software applications and incorporates artificial intelligence capabilities. Machine learning can be used to determine network peaks and solve the high energy consumption problem of 5G networks

At the recent Qingyun Cloud Yijie product communication meeting, Cao Zhen, Qingyun Technology Cloud Yijie product manager, also mentioned the ability of intelligent technology combined with Kubernetes AI to improve the efficiency of underlying computing power scheduling, thereby helping enterprises to create individual personalities Optimized AIGC application

It is not difficult to see that cloud native and AI have become two important technical forces in the digital transformation of the industry. However, for many companies, both cloud native and AI are technologies that are difficult to master and use for their own use.

For example, for a cement plant, due to its usually remote location, it also faces the problem of how to efficiently utilize power supply and achieve smooth operation and maintenance in the event of sudden power outage. This is similar to the energy-saving needs of the traditional telecommunications industry just mentioned. However, from the inherent personnel structure, it is difficult to find people who are proficient in cloud native or artificial intelligence technologies. However, at a cement group in Hunan Province, they recently adopted an easy and convenient solution, using cloud native technology to solve the past difficulties in IT architecture transformation, low operation and maintenance levels, and easy loss of business data in power outage scenarios. question

Shortly after the new version of Qingyun Cloud Yijie was launched, it achieved an easy and convenient transformation by dually enhancing cloud native and AI functions. This lightweight private cloud product is specially designed for small and medium-sized computing power to help customers fully unleash their potential in the digitalization process

Agility: Unstoppable with Cloud Native

According to statistics from Dynatrace, as cloud native tools continue to improve, auxiliary workloads now exceed application workloads, with a ratio of 63%:37%. This shows that Kubernetes is widely adopted by enterprises and that it is becoming more and more like the "operating system" of the cloud

How to better meet the needs of enterprises in the construction of container cloud platforms based on Kubernetes, not just for cloud native but cloud native, has become the key to how to design product architecture. The latest version of Cloud EasyJet architecture is divided into four layers. The bottom layer is based on Qingyun Cloud EasyJet's three most stable and reliable capabilities of computing, storage, and network to create Kubernetes clusters; the core layer can be built on the Cloud EasyJet platform. Quickly build Kubernetes clusters through visual operations; the basic functional layer has a functional suite of Kubernetes cluster management and application management. Through functional groups such as version management and resource management, it can manage a large number of container working clusters, and is connected to the underlying automation of Yunyijie. Scaling function; the application layer can support the transformation of business applications. The Qingyun container cluster is connected to the underlying cloudsat monitoring capability of Yunyijie, enabling cluster-level monitoring and alarms to understand the health of the container working cluster in real time.

"The characteristics of the Yunyijie container cluster are flexibility, simplicity, and freedom, which can effectively help enterprises complete the transformation of cloud-native applications." Cao Zhen summarized the advantages brought by the four-layer architecture.

Therefore, Yunyijie's natively integrated container cluster can ultimately realize the transformation of application architecture with containers as the core. For example, initial application containerization can standardize application delivery and environment. In application orchestration, complex applications can be deployed with one click, resource scheduling can be implemented, development environment inconsistencies can be resolved, and heterogeneous resources can be intelligently scheduled. In terms of application monitoring, efficient automated operation and maintenance is achieved. Yunyijie also provides a wealth of tools in the app store and provides process management for application development and testing

Intelligence: Give AI wings to traditional fields

In 2022, the scale of China's artificial intelligence industry will reach 195.8 billion yuan, with an annual growth rate of 7.8%. It is estimated that by 2027, the scale of the entire AI industry will reach 612.2 billion yuan. Although the domestic artificial intelligence industry is currently very hot, according to McKinsey's research, China's AI usage rate clearly lags behind other leading countries, and there is still a lot of room for improvement

For Yunyijie products, the Kubernetes cloud capabilities mentioned above have laid a good system foundation, and the rest is to make Kubernetes and AI hit it off. Therefore, Qingyun specially launched the Yunyijie AI all-in-one machine, which has built-in AI mirroring and Kubernetes container orchestration capabilities. It can package AI algorithms and applications into containers, and Kubernetes will automatically assign tasks to available computing resources to achieve resource allocation. Optimal utilization and efficient execution of tasks.

The bottom layer of this product has a hybrid computing cluster that supports artificial intelligence and rendering; the artificial intelligence computing scheduling layer supports dual management and scheduling of virtual machines and Kubernetes, and runs containers and virtual machines at the same time through the "dual engines" of containers and virtual machines. of heterogeneous workloads to ensure the provision of diverse computing capabilities

In response to the development needs of AI applications, Yunyijie integrates AI frameworks such as PyTorch, TensorFlow and MASS services. It has built-in GPT, GLM and LLaMA large models, and realizes AI application life cycle management through MCCFlow.

Yunyijie AI all-in-one machine has three main features: strong compatibility, high efficiency and multi-function, and data security. It supports x86 chips or Xinchuang chips, as well as domestic or imported GPU computing cards. In terms of large-scale models, it also supports domestic Tongyi Qianwen, Pangu, and Hunyuan, as well as foreign GPT-4, Claude, falcon, etc. The most important thing is that it can ensure that the business data of the training model is stored locally, and supports data encryption processing to ensure the security of business data

"Yunyijie" is a tool that everyone can use

"Yunyijie products should become a new generation of cloud infrastructure that everyone can install, use, and manage." Cao Zhen said when introducing the product positioning.

It is reported that simplicity and ease of use are the concepts that Yun Yijie products have always adhered to at the beginning of the design. The delivery environment insists on starting from three nodes, and the UI pages for deployment operations all implement visual operations. The interactive design is highly adapted to user habits and is equipped with live migration tools, which can be carried out without business downtime.

In terms of stability, the entire cluster adopts a master-node-less design, effectively preventing the risk of downtime of the entire cluster caused by a single node failure. It also supports online upgrades, capacity expansion, and migration operations to ensure that services can continue to run without interruption in the event of extreme downtime.

In the case of the cement plant mentioned above, the advantages of Yunyijie in simple operation and maintenance and stability are fully demonstrated. On the one hand, the high availability, hot migration, multiple copies and other functional features of Yunyijie's hyper-converged architecture are used to host key business systems and enhance the stable operation of business systems. On the other hand, the Yunyijie private cloud platform is connected to the UPS power monitoring system in the computer room. According to the policy set by the administrator, it effectively monitors the remaining power threshold of the UPS and automatically shuts down the virtualization and physical machines in the cluster to allow business data to be maintained during interruptions. It can land smoothly when powered, increasing data security

In response to the current needs of Xinchuang, represented by the financial industry, Xinchuang Cloud Easyjet products can help Xinchuang construction leaders in charge of Xinchuang Construction to solve Xinchuang hardware resource replacement and business system migration through Qingyun Technology's Xinchuang Cloud Easyjet, so as to Meet the task of completing localized replacement

Xinchuang Cloud Yijie focuses on core technologies such as domestic chips, operating systems, and cloud computing to quickly build a complete and easy-to-operate Xinchuang hyper-converged platform, which technically reduces the occurrence of loopholes and satisfies the needs of autonomy. Controllable requirements. The platform can adapt to hardware server chips such as Zhifeiteng, Haiguang, Kunpeng, and Zhaoxin, and supports operating systems such as Kirin, Tongxin, Hongqi, Deepin, Euler, etc., and has been completed with companies such as Renmin University of Finance and Economics, Dameng, and Dongfangtong. For adaptation testing, it also supports domestic office software such as Kingsoft WPS. Such comprehensive hardware, system, and software support solves the problem of poor compatibility between upstream and downstream products in the Xinchuang industry chain

It is not difficult to see that Qingyun’s new version of Yunyijie products not only provide customers with the tools to ride on the wave of technology: cloud native and AI, but also effectively integrate the two through platform integration. It is also easy to use and can quickly provide services for customers. Actual implementation scenarios bring digital rebirth. Let the enterprise's old systemization burden be turned into potential, and have the capabilities of cloud, ease and speed.

The above is the detailed content of Easily release potential and popularize cloud native and artificial intelligence through Qingyun Cloud. For more information, please follow other related articles on the PHP Chinese website!