According to Dynatrace, secondary workloads now outnumber application workloads with the continuous improvement of cloud-native tools, with a ratio of 63% to 37%. This also means that Kubernetes is already widely used by enterprises and is increasingly becoming the "operating system" of the cloud.
As a lightweight private cloud product designed for small and medium-sized computing power, the new version recently launched by Qingyun Technology Cloud Easy Jie has specially realized the dual enhancement of cloud native and AI, helping customers unleash the unlimited potential in the process of digitalization.
On the basis of Kubernetes, how to better meet the demands of enterprises in the construction of container cloud platform has become the key to how to design product architecture, not only for the sake of cloud native. The latest version of the Cloud Easy architecture is divided into four layers, and the underlying foundation is based on the three most stable and reliable capabilities of Qingyun Cloud Easy Computing, Storage, and Network to create Kubernetes clusters. The core layer can quickly build Kubernetes clusters through visual operations on the Cloud Easy platform; The basic functional layer has a suite of Kubernetes cluster management and application management functions, which can manage a large number of container working clusters through version management, resource management and other functional groups, and is connected to the automatic scaling function of the underlying layer of Yunyijie. The application layer can support the transformation of business applications, and the Qingyun container cluster is connected to the cloudat monitoring capability at the bottom layer of Yunyijie, which realizes cluster-level monitoring and alarming, and grasps the health of the container working cluster in real time.
The characteristics of Yunyijie container clusters are elasticity, simplicity, and freedom, so as to effectively help enterprises complete the transformation of application cloud native. Product Manager Zhen Cao summarized the advantages of the four-tier architecture.
Therefore, the natively integrated container clusters of Yunyijie can finally realize the transformation of application architecture with containers as the core. For example, application containerization at the beginning can standardize application delivery and environment, realize one-click deployment of complex applications in application orchestration, and realize resource scheduling, solve the inconsistency of development environments, and realize intelligent scheduling of heterogeneous resources. In terms of application monitoring, it implements efficient and automated O&M. Yunyijie also provides a wealth of tools in the app store and provides the best development and testing process management.
For the Cloud Easy products, the Kubernetes cloud capabilities described above have laid a good system foundation, and the rest is to make Kubernetes and AI implementation a perfect fit. As a result, Qingyun has specially launched the Cloud Easy AI all-in-one machine, which has built-in AI image and Kubernetes container orchestration capabilities, which can package AI algorithms and applications into containers, and Kubernetes automatically allocates tasks to available computing resources to achieve optimal utilization of resources and efficient execution of tasks.
The underlying layer of the product has a hybrid computing power cluster that supports AI or rendering; The AI computing power scheduling layer supports dual management and scheduling of virtual machines and Kubernetes, and uses the "dual engines" of containers and virtual machines to run heterogeneous workloads of containers and virtual machines at the same time to ensure the provision of diverse computing power.
The cloud easyjet AI all-in-one machine has three characteristics: strong compatibility, high efficiency and multi-function, and data security. Support x86 chip or Xinchuang chip, support domestic or imported GPU computing power card. In terms of large models, it also supports synonymous Qianwen, Pangu, and Mixed Yuan in China, as well as GPT-4, Claude, and Falcon in foreign countries. Finally, it also ensures that the business data of the training model is stored locally, and supports data encryption to ensure the security of business data.