Source | alibaba cloud native public number
Recently, Gartner, an internationally renowned information technology consulting agency, released the 2021 Container Competition Pattern report. Aliyun became the only Chinese enterprise selected for three consecutive years in China, and the product richness and maturity continue to maintain the global leading level.
Compared with previous years, on the basis of Kubernetes support, container image, Serverless container, service grid and other traditional dimensions, this report added two dimensions of cluster deployment form and management and control plane, Ali Cloud container products once again won high international recognition.
Gartner pointed out that containers have matured and entered the mainstream market, not only more abundant container enterprise landing scene emerged, but also a number of container manufacturers began to merge and integrate. Gartner said that to facilitate the adoption of containers in more mass industries, vendors need to continue to enrich the various enterprise application mirroring markets, increase support for aPaaS (Application Platform as a service), implement deployment and management of multiple cloud forms, and more.
The current product layout of Ali Cloud container services coincides with Gartner’s insights into the industry.
Ali Cloud launched a large number of new products for enterprise scale production of containers, container service enterprise ACK Pro and container image service enterprise ACR EE, to meet the strong demand for reliability and security; At the same time, it also provides corresponding functions in Serverless container, machine learning, edge computing and distributed cloud, and has been implemented in enterprises such as Internet, new retail, education, medical and other enterprises.
Serverless container, simple control, not simple massive capacity expansion
Serverless has become a hot topic of increasing concern. Its purpose is to help cloud applications reduce the workload unrelated to developing code, such as server operation and maintenance management, and focus more on business construction. At the same time, we will try our best to help customers optimize the application startup speed.
In addition to function computing (FC) for event-driven applications and Serverless Application Engine (SAE) for microservice applications, Ali Cloud also provides Serverless container products: Serverless Kubernetes (ASK), while providing the standard Kubernetes interface, not only allows users to enjoy the ultimate flexibility, and achieve o&M, Customers can quickly create Serverless clusters without managing Kubernetes nodes and servers, planning capacity carefully, or caring about the underlying servers.
The Pods in the Serverless Kubernetes cluster run in a secure isolated container running environment based on the Ali Cloud Elastic Container Instance (ECI). The bottom layer of each Pod container instance is completely isolated through lightweight virtual security sandbox technology, and container instances do not affect each other. The Serverless container architecture, which is deeply optimized based on Ali Cloud, enables ASK cluster to easily obtain great elastic capacity without being limited by the cluster node computing capacity. At present, the elastic capacity of ASK has reached 30 seconds to start 500 running PODS, and a single cluster can easily support 20K Pod capacity.
Yueguang Medical, a developer of mobile medical equipment, improved the prediction accuracy of AI model to 99.99% through Serverless container, while Weibo realized automatic business elastic scaling through Serverless container, reducing the end-to-end capacity expansion time by 70%.
Two cutting-edge application scenarios, machine learning (cloud native AI) and edge computing
Gartner believes that container technology and surrounding ecology are maturing and are capable of new application scenarios such as machine learning and edge computing.
1. Cloud native customization provides more optimization space for AI engineering efficiency
Ali Cloud container service has accumulated several years of cloud native AI practice scenarios, providing many tools, basic services and solutions in heterogeneous resource management, AI application life cycle management, AI task scheduling and elasticity, AI data management and acceleration, AI and big data unified orchestration and other directions. We have helped many customers improve the efficiency of AI production engineering, optimize GPU resource utilization, accelerate AI platform construction, etc. And promote the K8s community itself to expand to AI, big data and other large-scale data computing tasks.
Good future online education has deployed a large number of important business services under AI + education through Ali Cloud container services, and accumulated more than 100 AI capabilities covering the teaching links of “teaching, learning, testing, training and evaluation” and more than 10 AI solutions for education scenarios in image, voice and natural language processing. Not only improve the overall AI development and deployment efficiency, to achieve efficient MLOPS; At the same time, it solves the chronic problem of low resource reuse rate under the scenario of high concurrency, high performance and high resources of AI services, and the stability of platform ability invocation reaches 99.99%.
At the just concluded 2021 Ali Cloud Computing Summit, Ali Cloud launched the cloud native AI suite, which provides a variety of capabilities from bottom resource operation and optimization, multi-tenant quota management, AI task management and intelligent scheduling, big data service integration, to support the upper AI engine acceleration and domain algorithm application. At the same time, provide visual interface, command line tools and SDK, easy to use, easy to expand and integrate. AI service producers, including data scientists, AI algorithm engineers, AI platform builders and operators, are free to use the combination of optimization capabilities and choose to customize their AI platforms on Top of Kubernetes.
To apply for qualification for test, please click on the link to fill out the form: https://page.aliyun.com/form/act181046685/index.htm
2. Edge cloud native, which runs Kubernetes outside the cloud
The centralized storage and computing mode of traditional cloud computing center has been unable to meet the demand of edge devices for time-effectiveness, capacity and computing power, so how to build edge infrastructure has become a new topic.
The cloud native architecture of cloud side and side collaboration has emerged: in the center (cloud), the native cloud management and production capabilities are preserved, while a large number of edge nodes and edge businesses are transformed into workloads of the cloud native system. To realize the integration of services, o&M, and ecology, edge cloud native provides better service governance and traffic control, unified application management o&M experience, and better isolation, security, and efficiency.
Ali Cloud was the first to release edge computing container product ACK@Edge in 2019, featuring the service concept of “cloud standard management and control, edge moderate autonomy” : It provides powerful and rich cloud native management and control capabilities. It can realize the business integration of city brain, industrial brain, CDNPaaS and IoTPaaS upward, and complete cloud side operation and maintenance control through multi-specification and multi-link schemes carrying traffic down. Edge side, on the basis of the original K8s capabilities, superimpose such capabilities as edge autonomy, unitary management, traffic topology, edge computing state refinement detection and so on.
Sunac group builds a digitalized platform for community customer service based on ACK@edge, which manages two types of intelligent devices, such as gate system and information screen, and TV, intelligent air conditioner and Tmall Genie, in a unified manner. For example, it provides one-stop management for community parking lots to realize the intelligent collaboration of cloud side and end. Youku manages more than ten regions and numerous edge nodes on its public cloud in a unified manner through ACK@Edge, and reduces machine costs by more than 50% and end-to-end network latency by 75% through flexible expansion.
Distributed cloud breaks down boundaries, unified and finely controlled
Container management competition landscape will change significantly in 2020. Mainstream cloud service providers will release container management solutions of distributed cloud. Gartner predicts that 81% of enterprises will adopt multi-cloud/hybrid cloud strategies. “Most enterprises adopt multi-cloud strategies to avoid suppliers locking in or taking advantage of the best solutions.” Multi-cloud strategies achieve agility in enterprise procurement, balancing availability, performance, data sovereignty, and regulatory requirements and labor costs.
Ali Cloud container service ACK released hybrid cloud support in September 2019. In September 2020, ACK brought a complete upgrade of distributed cloud application management capability, which can not only manage ali Cloud K8s cluster, IT can also manage users’ own K8s clusters in IDC and other cloud K8s clusters to achieve unified cluster management, IT governance, security protection, application management and backup management. This year, Ali Cloud Container service also released a new unified application center delivery capability. Developers can release their applications safely, consistently and stably in multiple different cloud environments through GitOps, ensuring the whole process is safe and controllable, and improving the efficiency and stability of application delivery.
At the same time, Alibaba Cloud container service ACK can also be combined with managed service grid ASM, as the industry’s first ISTIO-compatible managed service grid platform. Service grid ASM provides unified capabilities for services running on different computing infrastructures, through traffic control, grid observation and communication security between services. Realize service nearby access, failover, gray release and other functions. Simplifies service governance across the board, applicable to clusters such as Kubernetes, Serverless Kubernetes, ECS VMS, and self-built clusters.
Ali cloud service grid ASM adhere to burnish “managed type, standardization, security and stability, easier to use, easy to expand,” the product features, has covered the 12 regional, continue to maintain product integrity, leading market share, not only internal bearing the industry’s largest grid cluster, in the service of alibaba group, focus on core business, such as electricity, nailing, etc., Moreover, it continues to lead in the scale of public cloud customers, covering online education/vocational training, e-commerce, logistics, supply chain, games, Internet of vehicles, IoT and other industries. Through service grid technology, service governance is decoupled from business, and sinks to the infrastructure layer, so that business development is more focused on business itself, and the multi-language multi-programming framework and various service governance challenges under microservice software architecture are solved in a systematic and standardized way. In super-large-scale application scenarios, while ensuring the stability of the system, it improves the observability, diagnosticability, perfect service governance ability and performance optimization through technical means.