Aliyun Jiangcen: The sublimation of cloud native in the form of edge

Introduction to May 20-22, the 13th China System Architect Conference (SACC2021) will be webcast live in the cloud, with the theme of "Digital Transformation, Remodeling of Architecture". Alibaba Cloud edge cloud native technology expert Jiang Cen shared Alibaba Cloud's exploration and practice in edge cloud native, and explained the core competitiveness of products from the aspects of coping with technical challenges and system architecture design, and driving business development with innovative technologies.

On May 20-22, the 13th China System Architects Conference (SACC2021) will be webcast live in the cloud, with the theme of "Digital Transformation, Architecture Reshaping". Alibaba Cloud edge cloud native technology expert Jiang Cen shared Alibaba Cloud's exploration and practice in edge cloud native, and explained the core competitiveness of products from the aspects of coping with technical challenges and system architecture design, and driving business development with innovative technologies.

Cloud Native Development and Status

With the maturity of cloud computing technology, most companies choose cloud computing to quickly deploy and operate their business. The large-scale commercial use of 5G will promote the networking of tens of billions of terminal devices worldwide. Customers' demand for near-end quasi-real-time computing with low latency and large bandwidth will greatly increase. The growth of the edge cloud computing market scale comes from the sinking edge of the central business on the one hand, and the emergence and development of various edge innovative business scenarios, such as cloud games and smart cities.

Jiang Cen believes that the enterprise business system going to the cloud, whether it is going to the central cloud or the edge cloud, will go through three stages:

The migration of self-built IDC, based on factors such as stability and disaster recovery, will not make major adjustments to the business structure, most of which only use the most basic cloud services, such as ECS, SLB, VPC, etc.;

For the overall business to go to the cloud, from the perspective of comprehensive cloud reuse capabilities and efficiency improvement and cost reduction, the evolution of the architecture born with the cloud has gradually begun to be applied in grayscale.

When everything is ready, the business begins to embrace cloud native on a large scale.

At this stage, many cloud services are already promoting cloud native on a large scale.

The concept of cloud native originated from the CNCF Cloud Native Computing Foundation and the Kubernetes platform incubated by Google. CNCF was established at the end of 2015 and has incubated a large number of high-quality projects that comply with cloud native standards. The core modules include databases, message middleware, application orchestration and scheduling, CICD continuous integration, RPC, service grid, container services, cloud native networks, and more.

Now, cloud native technology is no longer limited to the container/Kubernetes field, and has gradually become the standard architecture of the neutral software and hardware infrastructure of the majority of cloud vendors. Edge computing is a technology that has gradually emerged with the application of 5G and Internet of Things technology in the last 3-5 years. Its technology maturity is far lower than that of central cloud computing. Currently, there are not many projects involving edge computing on CNCF. With the improvement of edge scenarios and supporting capabilities, a large number of center businesses have sunk to the edge, and edge innovation scenarios continue to emerge, which will inevitably lead to the emergence of cloud-native technologies in line with edge characteristics at the edge.

Challenges in the native evolution of edge cloud

When talking about how cloud native technology evolves to the edge, Jiang Cen mentioned three technical challenges:

• From the resource side, the edge is different from the large-scale centralized layout of the center, which is mainly built with the goal of distributed and high geographical coverage. In addition to the central standard cloud server, there are also a large number of heterogeneous resources on the edge side, including Internet of Things equipment, MEC, cooperative joint construction of nodes, and so on. Cloud native technology has clear requirements for the deployment environment, so it needs to be flexibly adapted to the massive heterogeneous resources on the edge. In addition, the characteristics of edge nodes are small and large, and improving the resource reuse rate is the key. This requires flexible and elastic scheduling based on the ability of resource pooling and resource performance.
• From the perspective of technical capabilities, there are differences in cloud-side infrastructure. When cloud-native capabilities are directly applied to the edge, in addition to providing capabilities equivalent to the center's performance indicators, security isolation, disaster tolerance autonomy, and architecture awareness, they also need to be continuously improved The construction of high-speed passages at the edge of the cloud and the edge will increase the difficulty of construction.
• When resource adaptation and technical capabilities are available, maintaining a consistent user experience will face great challenges. From the perspective of users, the sinking process of the central business is bound to be a long process. For a single business center and the edge may coexist for a long time, the capacity building of the cloud edge is likely to be inconsistent, and most of the inconsistencies should be insensitive to users. Yes, so how to package products and achieve a consistent cloud-side experience in terms of cost, function, performance, and stability is extremely challenging.

Cloud Edge Cloud Native System Construction

Relying on 2800+ edge cloud nodes around the world, Alibaba Cloud provides users with secure, stable, and reliable edge computing and content distribution acceleration services, and builds edge cloud infrastructure closest to users. A single node is a small IDC with a scale ranging from a few to dozens of servers. The early edge cloud node establishment strategy was to establish a site separately from the CDN, which resulted in the inability to share resources and lack of business. The current construction strategy is to promote the integrated production of CDN ON ENS resources and integrate edge computing resources. After the integration, it will also bring greater possibilities for the time-sharing and reuse of resources.

As the most mature edge cloud application scenario, CDN has experienced long-term technological architecture evolution, and its infrastructure software and hardware architecture can be reused in edge cloud technology. The origin site is usually a server built by the enterprise itself, and its scale and performance are relatively limited compared to the central cloud. In the early stage of business launch, it can operate normally, but with the growth of business, in the face of massive client requests, if there is no CDN, the enterprise can only increase resource investment, otherwise it may cause the server's response timeout or even service paralysis. CDN, through multi-level caching and global DNS scheduling capabilities, enables users to access required resources (especially static resources such as pictures and videos) nearby, avoiding excessive pressure on source site bandwidth and servers. As it satisfies users from different regions to access nearby, it can be considered that CDN naturally has the typical characteristics of edge cloud computing with low latency and large global bandwidth. Supporting CDN monitoring, data intelligence, configuration management and other systems, with standard edge mass data distribution, processing, and the ability to interact with the center, it will gradually evolve into a supporting standard system native to the edge cloud.

According to the definition of the native capability model of Alibaba Cloud Edge Cloud: On the resource side, heterogeneous resources (including traditional physical machines, cloud-connected nodes, IoT/MEC equipment, ARM array servers, etc.) are pooled and clouded. It provides an edge cloud node operating system on it, virtualizes computing, storage, and network resources, and combines the container/K8S standard cloud native capabilities to build modular capabilities and extend the output community corresponding to the edge standard ecology. The ability of network-wide cluster application lifecycle management, orchestration and release corresponds to the ability of Alibaba Cloud to define the edge CRD operator EdgeWorkload to provide the ability to define OAM orchestration and expansion capabilities. For platform administrators, such as multi-cluster management, tenant isolation, metadata management, etc., it also requires corresponding capability customization in the scenario of massive users and massive data at the edge. In addition, there are a large number of distributed heterogeneous resources at the edge. How to maximize the use of resources requires a global container scheduler combined with business-related global traffic scheduling and distribution strategies. The flexible HPA/VPA scenario is also a distributed edge solution.

Alibaba Cloud has resources all over the world, and needs to define regional planning strategies for heterogeneous resource management modules, plan access, and develop around central control + edge autonomy + multiple caches.

Considering the complexity of the edge cloud architecture, the number of massive nodes, and the diversity of heterogeneous resources, Alibaba Cloud improves system stability by continuously improving system observability and strengthening Devops operation and maintenance capabilities.

At the same time, Alibaba Cloud Edge Cloud natively has technical advantages such as heterogeneous integration and wide coverage, cloud-side experience consistency, standard cloud native compatibility, and full-scale liquidity of computing power.

Typical edge cloud business application

The early CDN node architecture was mainly planned and deployed according to resources. 2 LVS+ is less than 4 control machines, and the rest are cache machines. This is a planning-first deployment mode, with more resources idle and a waste of construction costs. The comprehensive promotion of CDN ON ENS edge fusion computing can greatly improve the efficiency of resource utilization.

Smart terminal cloud access is a very important scenario for large-scale access of IoT devices in the future, involving the collaboration of typical edge global container scheduling and traffic scheduling. The central management and control will apply for resources in advance according to the estimated user scale, access the cluster, and deploy the container on the edge node. When the user requests to establish a connection, it will obtain the edge idle container from the central management and control according to the predefined traffic scheduling strategy. The user device and the server container are bound. When the user is disconnected, the new container is destroyed and rebuilt for subsequent use by other businesses to avoid data leakage. The central management and control will dynamically expand and shrink containers based on core indicators such as concurrent requests in real time.

For the center sinking business, the number of large-scale regions in the center is relatively limited. When customers are very sensitive to delay, the first choice is to deploy services at the nearest edge node and process customer requests. In order to ensure a consistent experience on the cloud side, the business central control system needs to obtain the service data of the center and the edge at the same time, and then distribute traffic according to user requests. This can not only reduce the pressure on the center bandwidth cost and resources, but also improve the user experience.

Finally, Jiang Cen said that Alibaba Cloud's edge cloud native technology will continue to improve its capabilities in scheduling, resources, and collaboration, providing industry customers and partners with the best cloud native application experience, and jointly creating innovative edge cloud applications.

Copyright Statement: content of this article is contributed spontaneously by Alibaba Cloud real-name registered users. The copyright belongs to the original author. The Alibaba Cloud Developer Community does not own its copyright and does not assume corresponding legal responsibilities. For specific rules, please refer to the "Alibaba Cloud Developer Community User Service Agreement" and the "Alibaba Cloud Developer Community Intellectual Property Protection Guidelines". If you find suspected plagiarism in this community, fill in the infringement complaint form to report it. Once verified, the community will immediately delete the suspected infringing content.