Research On Distributed Transaction Solution Under Kubernetes Container Cluster

With the rapid development of the Internet industry,the requirements for technology have become higher and higher.For example,12306 ticket grabbing software and Taobao Mall will have a concurrency of tens of millions or even higher at the peak.At the same time,a large number of data also inspect the storage capacity and search capacity of the system,and the requirements for high availability and concurrency of the system are increasing High.At the same time,software update iteration frequency is more and more frequent,micro service,Devops,continuous delivery,dynamic scaling and other concepts have been proposed and developed rapidly,and Cloud Native concept is more and more popular.Cloud Native does not simply transfer software services on traditional physical servers to cloud servers.It proposes changes and innovations from all aspects of application architecture,development,deployment,and maintenance,and ultimately brings the cloud's elasticity into full play,Dynamic scheduling,automatic scaling and other features.Cloud-native architecture and cloud-native applications involve many technologies,such as microservices,container technology,sustainable delivery,DevOps,and so on.This paper studies the main idea of cloud-native services as rapid service development and deployment through the Spring Cloud microservices framework,Docker containerization technology,and Kubernetes service orchestration technology.At the same time,combined with the container scenario,it solves the problem that the single-server transaction mechanism of a distributed system under multiple hosts cannot guarantee the idempotency of data consistency.First,this article analyzes the microservice system.Starting from the service split,the traditional single service is split by using the Spring Cloud framework.The gateway uses the Zuul gateway,the registration center uses Eureka,and the calls between microservices use the Feign component.Load balancing uses various Ribban policies in combination with various scenarios,and uses a combination of distributed sessions and global unique IDs based on snowflake algorithms to construct a microservice systemThe second part analyzes how to use Docker and Kubernetes.While microservices bring a series of benefits,they also bring some problems,such as the management and communication of multiple microservices,the deployment of multiple services,and the problems unavailability of other services due to problems with a single service.Here we use Docker to containerize services,to achieve efficient deployment of services,and to ensure the consistency of the environment,and to implement service orchestration,dynamic scaling and other functions through Kubernetes.In the third part,we discuss how to solve the distributed transaction problem of microservice in the containerized environment.Combining with several existing ways to practice and verify,we discuss the advantages and disadvantages and improve them.Finally,we choose the best scheme.