A few days ago, at the ICT China·2023 High-Level Forum-Cloud Native Industry Development Forum, the two major cases of "Shuhe Technology" and "Zhaolian Recruitment" provided technical support by Alibaba Cloud Container Service, as well as the cost of Alibaba Cloud's cloud- native AI suite and cloud-native FinOps The two major product technical solutions of the analysis suite jointly won the 2023 Cloud Native Application Practice Pioneer - Excellent Case of Cloud Native Technology Service.
Cloud native is an application development and deployment model based on technologies such as containerization and microservices. It can help enterprises quickly build and deploy applications, improve application reliability, elasticity and scalability, and enable enterprises to respond more agilely To meet market demand, quickly launch new products and services, improve the efficiency and speed of digital transformation, and use cloud-native technology architecture and concepts to build a new digital foundation are becoming the practical focus of digital innovation in all walks of life.
This time, Shuhe Technology and Zhaopin, which won the annual cloud native application excellent case award, made full use of the advanced nature of cloud native architecture to realize rapid deployment and update of applications in response to rapid changes in business. Scalability and reliability effectively guarantee the stability of the system under high concurrency and burst peak conditions.
Shuhe Technology AI Model Serverless Containerization
Driven by big data and technology, Shuhe Technology provides financial institutions with efficient smart retail financial solutions, serving licensed financial institutions such as banks, trusts, consumer finance companies, insurance companies, and small loan companies. Its business covers consumer credit, small and micro Enterprise credit, scenario installment and other fields, providing marketing customer acquisition, risk prevention and control, operation management and other services.
With the rapid development of business, the risk control model's demand for AI also blows out. How to create more agile AI computing services, support rapid business growth, and save resource costs at the same time is a major problem facing Shuhe Technology.
After research, Shuhe Technology and Alibaba Cloud's cloud-native team have cooperated in depth to jointly create a new Shuhe AI reasoning service platform. The entire platform is built on the basis of Alibaba Cloud's Serverless container service ASK, and the reasoning service container application can be directly deployed without purchasing any K8s nodes. There is also no need for node maintenance and capacity planning for the K8s cluster. At the same time, PODs are dynamically created and destroyed according to the real-time request volume of the inference service, which greatly reduces resource costs and greatly improves service availability. In addition, through the ASK Knative service, the problem of gray scale release and multi-version coexistence of the Shuhe model is effectively solved, and the efficiency of iterative online model is greatly improved.
At present, the system has deployed hundreds of AI model services and can provide hundreds of millions of query and decision-making services every day. ASK’s real-time elastic expansion capabilities help Shuhe save about 60% of resource costs, and at the same time smoothly cope with many operational activities. In the peak scenario, the entire cloud-native deployment and release system shortened the average deployment cycle of Shuhe from 1 day to 0.5 days, which effectively accelerated the iterative process of commercial applications and provided strong support for the expansion and growth of financial services .
Zhaopin.com upgraded to a cloud-native architecture to unleash the benefits of serverless technology
Zhaopin is a leader in China's human capital ecological value chain. Founded in 1994, it has served more than 11.76 million corporate customers and 321 million professional users, with 51.27 million monthly active users. As a human capital ecological value chain platform, it has a deep understanding and thinking about the characteristics and rules of the recruitment business. The application system is also updated and iterated along with the business development. During this period, it also experienced the slow upgrade of traditional applications, bloated architecture, and the stage of not being able to iterate quickly. .
At the beginning of 2018, Zhilian has started container transformation, and finally realized the ability of rapid business migration and expansion. In 2021, Zhilian continued to think deeply, based on its own business characteristics and combined with the advantages of cloud-native technology, it determined that serverless container hybrid cloud is the best technical solution for the further evolution of the system.
Zhaopin took the lead in the industry to complete a comprehensive containerization transformation, and jointly designed the "Distributed Cloud Container Platform ACK One + IDC Unified Scheduling Solution" with Alibaba Cloud. In response to the peak traffic phase during the spring recruitment period, through the unified cluster management of ACK One, the IDC container cluster combines cloud resources, unified observability capabilities, second-level elastic capabilities, and is compatible with offline scheduler strategies to achieve cost-optimized solutions and effectively solve During the spring recruitment period, we faced problems such as system stability, difficulty in estimating the amount of resources, and difficulty in operation and maintenance.
Using the hybrid cloud as a computing platform, there is no need to invest a large amount of capital at one time to purchase servers and expand cabinets due to sudden business growth. You can pay as you go on the public cloud, which is very convenient for some innovative businesses who want to do technical research, and pay as you go, which greatly reduces the cost of trial and error. In terms of efficiency, with the help of deep integration of the hybrid cloud architecture, it is convenient for R&D personnel to complete one-stop R&D, operation and maintenance work, and the entire continuous integration can be achieved in minutes.
The cloud-native transformation of Zhaopin has facilitated the rapid transformation of the company. The practice of cloud-native architecture and DevOps concept has made the development, testing, operation and maintenance of Zhaopin more collaborative and efficient. In addition, the high availability, flexibility, and automation features of the cloud-native architecture also enable Zhaopin to have better disaster recovery capabilities and business reliability, ensuring rapid business response and expansion during peak periods.
Cloud Native Technology Innovation, Serving Thousands of Industries
With the deepening of cloud use by enterprises, enterprises are faced with new problems of cloud cost management difficulties; at the same time, the complexity of Kubernetes is an important factor preventing many enterprises from adopting containers. Based on such pain points, Alibaba Cloud released the cloud-native FinOps cost analysis suite and cloud-native AI suite , which help enterprises realize cost visualization, optimization, and controllability through digital means and intelligent methods, and realize intelligent operation and maintenance, and quickly locate problems , improve operation and maintenance efficiency . These two product technical solutions also won the 2023 Cloud Native Technology Innovation Award.
The cloud-native AI suite **** focuses on continuously optimizing the efficiency of heterogeneous resources and efficiently running heterogeneous workloads such as AI in two core scenarios, providing unified management of heterogeneous resources, AI task scheduling, elastic AI tasks, AI data orchestration and access acceleration , AI job lifecycle management and other capabilities, providing AI/ML with an end-to-end solution with high engineering efficiency, low cost, scalability, and reproducibility.
At present, cloud-native AI kits have been widely used in the Internet, games, autonomous driving, quantitative finance, online education, AI for Science, scientific research institutions and other industries; K8s GPU shared scheduler, cloud-native data orchestration and acceleration solution, and AI task lifecycle management tool Arena, work together with academia and the community to promote the evolution and implementation of cloud-native AI, and continue to provide value to more enterprises and developers.
We know that compared with traditional applications, container applications have higher elasticity and dynamics, and can create and release resources on demand, but at the same time, this also brings greater challenges to cost estimation.
The cloud-native FinOps cost analysis suite provides the cost insight function, which precipitates a set of mature IT enterprise cost governance processes and systems, and reduces the IT cost governance cycle from the original quarter and month to days and hours ; at the same time, the cost insight function provides Through the logical management of multi-tenancy business, financial asset splitting, and waste measurement for the application dimensions of business such as namespace and workload, it is possible to easily realize the cost allocation of different services in a K8s cluster, and realize capacity management and financial asset management. And it provides a cost allocation model based on different resources such as CPU, memory, GPU, etc.
At present, the cloud-native FinOps cost analysis suite has served a large number of enterprise customers. Enterprises such as China Property Insurance and TCL have saved millions of costs every year through cloud-native enterprise IT cost management solutions and product capabilities.
With more than ten years of container technology reserves and practical experience, Alibaba Cloud Container Service has the ability to support super-large clusters, and is the first in China to achieve a breakthrough in the scale of 10,000 nodes and 1 million Pods in a single cluster, which can help enterprises easily cope with the ever-increasing scale needs. In the public cloud container platform report released by Forrester, it was rated as the market leader of the public cloud container platform.
Ding Yu, head of Alibaba Cloud's intelligent cloud-native application platform, said that Alibaba Cloud continues to expand the capabilities and boundaries of the cloud through a comprehensive container product layout. At the infrastructure layer, new computing power oriented to cloud-native optimization is used to improve computing efficiency; at the container orchestration layer, applications are made more stable through capabilities such as elasticity and mixed-part scheduling; at the application architecture layer, technologies such as service grid are used to Improve application resilience and scalability. In the field of operational governance, use the FinOps suite to reduce resource idleness.
Cloud native has become the technical force behind digital innovation in thousands of industries, and containers have also become a new interface for cloud computing. In the future, Alibaba Cloud Container Service will continue to be more intelligent, making enterprise innovation within easy reach.
At present, the serverless container service ASK has joined the Feitian free trial plan, click here to receive free trial resources.