Research On Service Discovery And Service Reliability Based On Microservices Architecture

A system based on microservices architecture consists of many small services that communicate with each other over the network.Each service has a separate software development life cycle,but they rely on each other.Although there are many related researches on microservices architecture,there are some problems.Including,service discovery delay is too large,accuracy is not enough,service quality description of single service is not comprehensive,service in dynamic environment always fail,and they can not adjust themselves instantaneously.In the modeling research of microservices architecture,the current dependency graph generation scheme is not convenient,and its accuracy is low.The current fault trees model does not take fault tolerance and the actual situation of the system into consideration.In the study of service risk,the existing risk calculation model does not take the effect of different execution path into consideration.This thesis mainly studies service reliability,service discovery,system reliability modeling,service risk and service recovery in systems based on microservices architecture.The results are summarized as follows:The reliability of service in systems based on microservices architecture needs to be studied.In this thesis,QoS calculation model is proposed for the calculation of service quality,and it can be used to evaluate the service running state.Based on the service running state,an adaptive service model is proposed to achieve the goal of service fault tolerance and throughput improvement during service running.The experimental results show that QoS computing model and adaptive service model proposed in this thesis increase the success rate of the service in dynamic environment by 3.87%,and increase throughput by 18.7%.This thesis studies the problem of service discovery in microservices architecture,how to model dependencies between applications,and how to use these models to analyze the reliability of systems based on microservices architecture.An automated generation scheme of dependency graph is proposed and fault trees model is improved.And they are used as methods to analyze the reliability of the system.The experimental results show that fault trees model can effectively analyze the reliability of systems based on microservices architecture.The improved fault trees model is used to analyze the fault rate of the system,the accuracy has been improved to a certain extent,and the accuracy in the experimental environment is improved by 4.53%.Services always deploy independently.How to efficiently monitor and manage these services is an important issue.If we just solve the service failure during deploying or running,it will lead to a large number of failures.Therefore,the prediction and recovery of service failures is equally important.In this thesis,a fault recovery scheme based on risk is proposed.The removal of risky services can be completed before a large number of failures occur,so as to avoid the occurrence of continuous errors.The experimental results show that the proposed risk calculation model and fault recovery scheme can improve the robustness of system,and can reduce the waste of resources caused by fault recovery scheme.