Alibaba Cloud Container Service adds support for the Kubernetes orchestration system, significantly improving performance

Scenario Study 2017-11-29 10:33:53 Views 90 Comments 0
Deep Learning Architecture Docker Performance Public Cloud Cloud Server Image Cluster Microservice Container Elastic Scaling Load Balancing Distributed Storage Enterprise Application kubernetes

Abstract: As two major schools of container orchestration systems, The importance of Kubernetes and Swarm cannot be overstated. Alibaba Cloud Container Service, which integrates two high-performance integrated services, can not only reduce infrastructure costs by 50%, improve delivery speed and accelerate product iteration by 13 times, but also enable massive container startup in seconds, application architecture scaling and Recovery, deployment in minutes.

Alibaba Cloud Container Service provides technical capabilities for enterprise customers, and provides migration tools and consulting services, deep learning, blockchain and other application solutions for enterprise application containerization to help enterprises optimize existing IT investments and accelerate business innovation.

At the Yunqi Conference on October 12, the new version of Alibaba Cloud's proprietary cloud (Feitian Proprietary Cloud Agile Version 2.0) announced support for Kubernetes; a few days ago, the public cloud container service also announced that it will support the hosting of Kubernetes version 1.8.1. service, and became one of the first manufacturers in the world to pass the Kubernetes conformance certification. As the two major schools of container orchestration systems, the importance of Kubernetes and Swarm is self-evident. Alibaba Cloud Container Service, which integrates two high-performance integrated services, can not only reduce infrastructure costs by 50%, improve delivery speed and accelerate product iteration by 13 times, but also enable massive container startup in seconds, application architecture scaling and Recovery, deployment in minutes.

Project Background

In the case of enterprise-level large-scale containerization, how to deploy applications in a distributed environment, how to manage cross-machine applications, how to maintain and implement load balancing, resource quotas, automatic scheduling, online capacity expansion, etc.: This is container orchestration the role of the system. The English word of the container orchestration system is container orchestration, in which orchestration is literally translated as "orchestral arrangement". When arranging, it is necessary to consider how to interweave different instruments and how to make the music more beautiful through different movements.

Enterprises need a container orchestration system when managing container clusters. At present, the two mainstream solutions are Kubernetes from Google and Docker Swarm developed by Docker. Among them, Kubernetes is a cluster management software for automatic deployment, expansion and management of containerized applications. It supports a series of container tools including Docker.

As the only domestic partner of Docker and a CNCF gold member, Alibaba Cloud has been focusing on providing container services for enterprises. At the end of 2015, the public beta of Alibaba Cloud Container Service was launched for the first time; it was officially commercialized in May 2016; in October 2016, it became the only domestic partner of Docker and launched the private cloud enterprise version. In December, the product was internationalized. The original container service supported the Docker Swarm orchestration system, and in May 2017, support for the Kubernetes orchestration system was added.

Technology Interpretation

The biggest highlight of this product upgrade is the high integration of container technology and cloud capabilities, providing stable core capabilities of container cloud. The Cloud Provider Controller provided by Alibaba Cloud realizes the seamless integration of native Kubernetes and Alibaba Cloud capabilities. It can easily use Alibaba Cloud's powerful resources, including cloud hosts, load balancing, distributed storage, and heterogeneous computing. The IaaS capability has carried out multiple optimizations on Kubernetes, providing users with a safe, stable and easy-to-use Kubernetes managed service.

Alibaba Cloud will demonstrate 6 unique advantages:

simple load balancing: SLB can be easily used as the load balancing entry for applications;
efficient container network interface: Kubernetes relies on network plug-ins to achieve interconnection between containers. The container service team contributed a CNI-compatible Alibaba Cloud VPC network model to the open source network solution flannel. The network model is simple and efficient, and can support mutual access between containers and hosts on the same network plane, which is very suitable for hybrid cloud deployment;
Persistent file volume support: When a virtual machine node fails, Kubernetes will migrate containers to other hosts. Migration of data is not possible with locally stored containers. Alibaba Cloud Container Service adds support for distributed storage services such as Alibaba Cloud Disk and NAS to Kubernetes. In this way, when the container is migrated to other nodes, the cluster nodes will automatically unbind and bind the corresponding storage for the container to ensure that the application can be migrated and restored normally;
GPU, heterogeneous computing and deep learning: When creating a Kubernetes cluster, not only can Select CPU worker nodes, and optionally GPU-enabled worker nodes. GPU resource scheduling is also enabled in the cluster, enabling heterogeneous computing clusters to support deep learning and other scenarios;
stable version and official certification: At the same time, in order to optimize the stability of the container runtime and support for new features, the container service updates Docker to the stable community version 17.06.2. Alibaba Cloud Container Service has passed CNCF's Kubernetes conformance test and can provide customers with a compatible and stable container running platform;
one-click deployment, DevOps integration: Kubernetes deployment and operation and maintenance are very challenging for entry-level customers. Alibaba Cloud Container Service further improves ease of use and lowers the threshold for deployment, management, and application development. Container Service supports one-click deployment of Kubernetes clusters on Alibaba Cloud, and integrates with the Kubernetes console. Users can manage clusters through Alibaba Cloud accounts. What's more convenient is the built-in integration of Kubernetes solutions in Alibaba Cloud CodePipeline, which can greatly simplify the process from code submission to launch.
In addition, the container image service, one of the products provided by Alibaba Cloud Container Service, has also entered the public beta stage. The mirror service has been verified by Alibaba Group on Double Eleven, and has supported more than 100,000 mirror warehouses and 200 million mirror downloads. Alibaba Cloud Container Image Service provides secure image hosting capabilities. It is deployed in more than 12 regions around the world. It can provide stable image construction services, convenient team organization and collaboration functions, etc., to facilitate users to manage the full life cycle of images.

Scenario application

The following is the product architecture diagram of Alibaba Cloud Container Service. The three application scenarios of DevOps, microservices and enterprise applications mentioned at the top level, and more detailed ones include machine learning, elastic scaling and hybrid cloud scenarios (click here to view Schematic diagrams of the five actual combat scenarios).

d7898807dabde28262d86263a7dc62dd56243548 The

early Kubernetes was not very mature, with weak security capabilities and complex deployment. However, Kubernetes developed rapidly in 2016 and is now very complete. Therefore, the Alibaba Cloud container team chose to launch the Kubernetes service at the right time, and added support for Kubernetes on the basis of the previous Docker Swarm. In this way, users can choose different technical solutions according to their needs.

All roads lead to Rome. Both are good orchestration systems, focusing on how to use container technology to help enterprises innovate. Whether it is Swarm, Kubernetes or its own solution, users can choose an orchestration solution according to their own situation, and have greater freedom in technology selection.

It is reported that Alibaba Cloud Container Service has helped many domestic and foreign customers to realize application migration, DevOps, micro-service architecture, AI business innovation, hybrid cloud, elastic scaling and many other scenarios in Alibaba Cloud public cloud and proprietary cloud. Business/O2O, finance, games, big data, education, media, industrial manufacturing, cloud computing, medical and health industries and other enterprises provide services, such as Easyhome, Shanghai Stock Exchange, CITIC Group, ofo bike sharing, BGI Wait.

Practical access

In order to lower the threshold for development and application, Alibaba Cloud has supplemented Kubernetes capabilities in multiple ways, such as implementing heterogeneous computing clusters to support deep learning and other scenarios by selecting different nodes, or deploying cluster integration solutions on the cloud with one click.

Some articles are convenient for you to quickly start using container services:

"Create a Python Application Using Redis on Alibaba Cloud Container Service"
"Using Docker and Alibaba Cloud Container Service to Easily Build a TensorFlow Serving Cluster"
"When Docker Meets Databases: Using RDS in Alibaba Cloud Container Service"
"Docker-based Application Load Balancing and Service Discovery"
"Using Alibaba Cloud Container Service to Implement Customization" For
more information on the operation steps of the container image service, please refer to "Four Questions to Understand Alibaba Cloud's Image Repository and How to Perform Full Lifecycle Management for Images?" "In the article - What is the example of the use of mirror warehouses?

As of now, Alibaba Cloud has become the only Chinese cloud computing company in China that supports Kubernetes and Swarm, two popular container scheduling systems. The public cloud version of Alibaba Cloud Container Service is free, but the use of the private cloud version is paid. The container service fee includes two parts: the cost of the resources used (cloud servers, load balancing, etc.) and the management fee of the container cluster . Cloud server instances or load balancing instances that are automatically created or manually added through Container Service are still billed at the relevant cloud server and load balancing prices. For specific billing information, see Cloud Server Billing and Load Balancing Billing.

Copyright statement: The content of this article is contributed by Internet users voluntarily, and this community does not own the ownership and does not assume relevant legal responsibilities. If you find any content suspected of plagiarism in this community, please send an email to: [email protected] to report and provide relevant evidence. Once verified, this community will immediately delete the allegedly infringing content.