Research On Reliable-oriented Adapation On Microservice System

The microservice architecture is gaining popularity as an architecture style to achieve extreme agility.One of the biggest “complaints” against the microsevice architectural application is that while we gain agility from the microservice architecture style,gaining the reliability insight about the entire operation of the microservice architectural sytem becomes more difficult.In this paper,the key problems of reliable-oriented adapation on microservice architectural system are focused.Firstly,service quality prediction plays a very important role in many fields related of service-oriented systems.Now,there are large numbers of approaches to predict the quality of service-oriented systems.However,it is impossible to apply one method to all prediction tasks without distinction.In reverse,unclearly to recognize the category of the prediction tasks will result in losing in all kinds of prediction approaches.In order to discover which methods are suitable for their corresponding service quality prediction scenarios,it is necessary to classify the prediction tasks and to classify the prediction approaches clearly.In this paper,a classification framework for predicting tasks for QoS of service-oriented systems is defined,four types of the prediction technologies are mainly introduced,and the research findings of quality prediction approaches for service-oriented system in literature using those four types of prediction technologies are listed and the increasing use of the method based on probabilistic model checking are specially concentrated on,and finally the result of this survey are given and the open research questions about predicting service quality by probabilistic model checking method are discussed.Secondly,conventional monolithic architecture multi instance deployment mechanism in response to large traffic,high concurrent,flood like users access,effectively alleviate the pressure of the system,while resulted in a great waste of resources.Microservices architecture for its not only alleviating the system pressure but also making full use of the system resources becomes the preferred system architecture.Reliability,as a quality attribute of software system of the concern of stakeholders,is also a key quality indicator of microservice architecture system.The previous reliability analysis methods are for the monolithic architecture system while they can not be used directly to analyze the reliability of microservices architecture system.To deal with system reliability analysis under the microservices multi-instance deployment mechanism,this paper presents a method of microservices architecture system reliability analysis based on big data.And this method is demonstated by experiment.Thirdly,the traditional solutions to the reliable service flow combination problems all mark the same service and service model granularity as the precondition.In fact,there is a class of large-grained service on the Internet,that is,a service can complete the requirement specification of multiple service models.Compared with the service flow composition with services whose granularity are the same as the service models,the using of large granularity services to composite service flow improves the efficiency of the combined process by reducing the number of candidate services set to participate in the composition process.This paper is devoted to the research on reliability-awared service flow combination with large granularity services.Firstly,based on quotient space granular computing model,this paper illustrates that service flow combination problem based on the small granularity service model can be converted into the service flow combination problem that based on larger one;then points out two problems that need to be sloved,one is to construct the best large-grained service process combination model,another is the using of the large grained services.The effectiveness of the proposed solution in this paper is demonstated through the experiment.Finally,the paper develops a tool of reliable-oriented adapation platform on microservice system using the above all techniques proposed in this paper.By using the tool,the microservice system can be adapated according to its reliability.