Research And Application Of End-to-end QoS Aggregation And Container Allocation Optimization For Microservice Flow

With the rapid increase of business volume in recent years,the problems in scalability and usability of single architecture are highlighted.Once the program functions are changed,the whole application needs to be rebuilt and deployed.Microservice,as a new application development model,has attracted wide attention due to its advantages such as independence,scalability,and decentralization.Each microservice can be implemented,deployed,and updated independently without affecting the integrity of the application.As a lightweight virtualization technology,the container supports the encapsulation and deployment of microservice applications.Container technology provides convenience for the execution and deployment of microservices and are one of the important technologies to support microservices.A microservice flow composed of a set of microservices can be represented by a complex request execution path that spans multiple microservices.Microservice flow greatly improves the scalability,portability,updateability,and availability of applications.However,the cost is high overhead and complex resource usage dynamics.It is essential to aggregate the quality of service(QoS)of single microservice to provide comprehensive QoS measurement for the end-to-end microservice flow.It is irrational to blindly allocate resources to applications to meet increasing workloads.Service providers want to minimize the resources allocated to applications to reduce operating costs,and ensure that the resources provided can support the service level agreement(SLA)reached by users,they must develop a reasonable container resource allocation strategy.At present,there is still a lack of research on the container allocation of microservice flow in the microservice field.In addition,it is also a challenge to find the optimal end-to-end container allocation scheme with QoS guarantee for microservice flow.The main work of this paper is as follows:Firstly,through the investigation and analysis of related literature,this paper introduces the background of microservice architecture and resource allocation,as well as the domestic and foreign research status of microservice and resource allocation,and then the main research work of this article is given;Secondly,we define a microservice flow model,and then describe the relationship among microservices as four combinations based on the QoS aggregation method in Web service.On this basis,we formulate the end-to-end QoS aggregation model of microservice flow;Thirdly,to ensure reasonable container allocation among microservices,an end-to-end container allocation optimization model for microservice flow is proposed,which is represented as a nonlinear optimization problem and solved by the proposed ONSGA2-DE algorithm;Fourth,a comprehensive evaluation of our modeling method and optimization algorithm on the open-source microservice benchmark Sock Shop.Experimental results show that our method can reasonably allocate containers for microservice flows under the premise of guaranteeing QoS;Lastly,based on the Chang'an Travel project,the model of mobility business microservice flow is established.Then we design a complete business flowchart using the Spring Cloud and optimize container allocation for the mobility microservice flow through our proposed model.