Containers Are the Shortest Path to Unlocking Cloud Value

Date: Nov 1, 2022

Cloud has changed the IT format and market structure, and has given birth to an era of great application development. Enterprises can focus more on building more viable business innovations that meet their vision. The era of comprehensive use of cloud services to build software has come. In this context, the concept of cloud native has been proposed and quickly visualized, and cloud native technologies represented by containers, as the best choice to improve cloud service capabilities, are also developed rapidly.



In January 2019, Tripwire released its latest state-of-the-art research on containers in 2019. According to the feedback, up to 86% of the respondents use containers in production environments, and nearly 32% of respondents use more than 100 containers in the production environment of their enterprises. Overall, the use of containers in production environments has been greatly improved. Survey forecasts show that the current application container market will grow from $762 million in 2016 to $2.7 billion in 2020. It is not difficult to see that the tremendous flexibility exhibited by the introduction of containers has effectively pushed the pace of its adoption, making enterprises increasingly reliant on the technology.



In recent years, container technology and related applications have received more and more attention at home and abroad, and the development momentum of R&D and application promotion is rapid. In this issue of "The Evolution of China's IT Technology", we had the honor to interview Ding Yu (Hua Ming Shutong), the head of Alibaba Cloud's intelligent container platform, about the development path of container technology in the cloud computing era, and Alibaba Cloud's product matrix in the container field , technology iteration and future trends, analyzes the evolution of Alibaba Cloud's container technology, and analyzes the development trend of the entire industry.

Cloud-native technologies represented by containers have become the shortest path to unlock cloud value in the cloud era

"In the past, we often used virtualization as the cloud platform and the interface for interacting with customers, which brought flexibility to the enterprise and also brought a certain management complexity; It has gradually become one of the new interfaces for cloud platform and interaction with customers. The construction, distribution and delivery of applications can be standardized at this level, which greatly reduces the cost of enterprise IT implementation and operation and maintenance, and improves the efficiency of business innovation. In terms of dimensions, open source makes cloud computing more and more standardized. Containers have become a standard for application distribution and delivery, which can decouple applications from the underlying operating environment; Kubernetes has become a standard for resource scheduling and orchestration, shielding the differences in the underlying architecture. It helps applications run smoothly on different infrastructures; the upper-layer application abstractions established on this basis, such as microservices and service grids, gradually form the standard for the modernization and evolution of application architecture. Developers only need to focus on their own business logic, no need Focusing on the underlying implementation, cloud native is reshaping the entire software technology stack and life cycle through methodology, toolsets and concepts. The cloud native technology represented by containers uses an open and standard technology system to help enterprises and developers build on the cloud And running a system that is elastically scalable, fault-tolerant, easy to manage, and easy to observe has become the shortest path to unlock the value of the cloud." Uncle Tong emphasized when referring to the evolution of containers. Internet companies were the first to create and apply container technology. Today, an open-standard cloud-native ecosystem has made container technology popular. More and more enterprises and developers use containers to build applications and share the benefits of this technology.



What new requirements do enterprise users have for container technology? In this regard, Uncle Tong said that in the process of landing cloud native applications, security is one of the most concerned requirements of enterprise users. The traditional RunC container shares the kernel with the host Linux, and provides limited isolation through CGroup and namespace. As more and more enterprise customers begin to pay attention to container security, new types of highly isolated and secure runtimes have emerged in the past two years, including MircoVM (Kata Container, FireCracker) orientation and gVisor security sandbox orientation. Alibaba Cloud and the Ant Financial team cooperated to introduce the security sandbox container technology, and released the RunV security sandbox based on lightweight virtualization technology in September 2019. Compared with the RunC container, each RunV container has an independent kernel, even if If the kernel to which the container belongs is compromised, it will not affect other containers. Alibaba Cloud Container Service provides an end-to-end cloud-native security architecture, including infrastructure security, software supply chain security, and runtime security, providing enterprises with comprehensive, three-dimensional, and multi-layered security protection.

The second requirement is the hybrid cloud architecture. Going to the cloud is the general trend. Many customers will consider the hybrid cloud method due to business reasons. The differences in the infrastructure and security architecture of different cloud environments will cause the separation of enterprise IT and operation and maintenance systems. Manage complexity. In the cloud-native era, technologies represented by containers and Kubernetes shield infrastructure differences, and serve as the base to promote the arrival of the application-centric hybrid cloud 2.0 architecture to meet user demands.

At the same time, enterprises have also put forward higher requirements in terms of operational efficiency, R&D efficiency, operating costs, system fault tolerance, and maintainability. Alibaba Cloud is committed to solving the pain points of enterprises during the entire development process of container products. Although enterprises have different requirements for cloud migration, containers and cloud native, as a universal technology, can meet the needs of different enterprises at different levels.



Talking about Alibaba Cloud's innovation in container technology and products, Shutong emphasized, "Alibaba Cloud hopes to continue to follow the community route in the cloud native field, be fully compatible with open source technologies, and take advantage of the scale and complexity of Alibaba's economy and the rich scenarios of Alibaba Cloud customers. In the beginning of 2019, we settled the best practices of cloud-native mass production and opened it up to the community in the form of an open source project OpenKruise, on the one hand, to help enterprise customers In the process of cloud native exploration, avoid detours and reduce technology fragmentation. On the one hand, it promotes the upstream community to gradually improve and enrich application automation management capabilities; in October 2019, Alibaba Cloud and Microsoft jointly released the Open Application Model (OAM). ), OAM is a standard specification that focuses on describing the application life cycle, which can help application developers, application operation and maintenance personnel and infrastructure operation and maintenance teams to better collaborate. In this model, developers are responsible for defining application composition, dependencies and architecture. ; Application operation and maintenance personnel are responsible for defining application runtime configuration and operation and maintenance requirements, such as release policies and monitoring indicators, while the infrastructure operation and maintenance team can configure customized parameters for different application deployment environments. Through this separation of concerns design , which can decouple application definition, operation and maintenance capabilities and infrastructure, make application delivery more efficient, reliable and automated, and solve industry pain points. Serverless, through a higher level of abstraction, frees developers from resource management and daily operation. Dimension and other work, to achieve simplified R&D, extreme flexibility, and pay-per-use. Alibaba Cloud has created a function computing product, FunctionCompute, which provides event-driven programming, and provides an application-oriented serverless application hosting platform SAE. Users only need to provide application implementation, The platform is responsible for elastic and automated operation and maintenance; it provides container-oriented serverless products ECI and serverless Kubernetes. On the other hand, it is also promoting the serverlessization of some traditional technologies, such as databases and message middleware. Driven by technological innovation, Alibaba Cloud hopes to Become the best cloud native product implementation, continue to leverage the advantages of the largest cloud native application practice in China, the most abundant cloud native product family in China, the largest cloud native customer group in China, and the most comprehensive cloud native open source contribution in China. A wide range of enterprise customers and developers."



The evolution and upgrade of the cloud-based architecture during the 11-year Double 11, creating the best innovation soil for container technology



"The biggest difference between Alibaba Cloud and other cloud vendors is that Alibaba's core business runs on the cloud, which forms the best soil for innovation. That is to say, our most advanced technologies will first be tried in Alibaba's own business system. It has been used on a large scale, proving the universality and value of its technology before opening it to customers." When talking about the progress of Alibaba Cloud's containerization, Shu Tong emphasized that any technology will be tried and matched in Alibaba's own business system. Application, in 2011, Alibaba Cloud began to step into the container door. In 2013, when Docker came out, Alibaba Cloud Container quickly integrated its advanced concepts, and in 2015, it promoted the comprehensive containerization evolution of the group's business, and this series of development and evolution are actually inseparable. To meet the needs of the Double 11 promotion, for example, comprehensive containerization can help the Double 11 promotion to achieve rapid and elastic expansion.



In the experience of the Double 11 event, millions of containers supported the smooth progress of the Double 11 event. Due to the extremely large scale of the business, its complexity is very high, which also brings greater challenges to container technology. For example, in the process of container image distribution, tens of thousands of images are released and distributed at a time, and such huge traffic is not a small challenge. In order to achieve the ultimate requirement of efficiency, Alibaba Cloud uses P2P technology to realize large-scale and large-scale rapid distribution, and realizes the completion of cross-machine room image download and container startup within 10 seconds.



The significant impact of container technology on Double 11 is also included in the specific implementation of co-location technology. Shutong said that through co-location technology, Alibaba Group can save about 30% of IT costs, which can be used in the special time period of Double 11. , reducing the cost per 10,000 transactions by more than 75%.

Containers, Microservices, AI Future Trends: Collaboration, Convergence

Container technology has been widely recognized by the industry, and its future development prospects not only depend on its outstanding performance in the technical field, but also need to be integrated with more technologies to become a successful product technology that progresses with the times. In the early days, Kubernetes mainly ran stateless web applications, such as microservice applications based on Apache Dubbo/Spring Cloud. Now, more and more enterprise core business, data intelligence business and innovative business are also running on Kubernetes. Taking Alibaba Cloud's own cloud products as an example, including enterprise-level distributed application service EDAS, real-time computing platform Flink, elastic AI algorithm service EAS, and blockchain platform BaaS are all deployed on Alibaba Cloud's Kubernetes service ACK.



From the perspective of application architecture evolution, the development of containers has promoted the development of microservices. The big problem encountered in the early implementation of microservices is the complexity of operation and maintenance and the inconsistency of the environment caused by the split of the architecture. Microservices have grown rapidly. The introduction of the microservice architecture solves some problems, at the same time, it invades the R&D framework, and the framework iteration and R&D iteration are coupled, and the support in the multi-language environment is not friendly enough, and the management is more complicated. Therefore, the community began to try Service Mesh, gradually sinking microservice capabilities from framework capabilities to platform capabilities. It can be seen that containers and microservices are promoting each other.



"Cloud native and AI are excellent partners, and the two are mutually empowering." Uncle Tong emphasized when referring to the integration of AI and containers. First of all, AI is an emerging field, and there are not so many historical burdens in the architecture. In addition, AI computing itself has requirements for flexibility, resource efficiency, and deployment efficiency. Container technology can solve the above problems. New architectures such as GPUs, FPGAs, and proprietary ASIC chips bring huge increases in computing power and difficulty in management and maintenance. Using Kubernetes to provide unified management and efficient scheduling of heterogeneous resources, improve flexibility and support fine-grained sharing. It can improve resource utilization by 3 to 5 times. AI is also of great help to container cloud-native technologies. AI often represents business scenarios, which provides a rich space for verification of how cloud-native technologies are more universal, thereby improving the maturity of cloud-native technologies.



It has been more than 6 years since the emergence of container technology, and the rapid development of Kubernetes is not news, but this does not mean that the ecosystem of container technology has developed smoothly. On the contrary, the technical system of the container and its surrounding is still maintaining rapid development. Talking about new technologies and new directions that will be of interest in the future, Uncle Tong admitted that containers should go to all environments, not only traditional IDCs, but also public clouds, proprietary clouds, edge nodes, Internet of Things, big data, AI In various scenarios, we hope to use cloud native technology to lower the threshold of cloud computing and truly realize the inflection point on the cloud.