Resarch On Extraction Of Microservices In The Process Of Legacy System Refactoring

With the passage of time and the continuous and rapid evolution of software system requirements,the scale of the application continues to expand,and the application once developed and maintained by a team has become a huge monolithic system.Monolithic systems cannot meet rapidly changing business requirements and lead to slow development speed,extended deployment cycles,difficulty in scaling and reliable delivery,and inability to adapt to new technology stacks.In response to these problems,many organizations restructured monolithic legacy systems into microservice architectures system.Domain Driven Design(DDD)defines a separate domain model and bounded context for each subdomain,and defines the subdomain as a service,and the bounded context determines the granularity for the service as the boundary of the service.Based on the relationship between the domain model and the service,the paper proposes a method for extracting microservices based on the domain division of the legacy system model.The domain model division is based on latent semantic analysis(LSA).First,establish the LSA model and calculate the cosine similarity between the class documents.The class documents are used as nodes.If the similarity value between the class documents is within the range of the number of topics in the LSA model,an edge is established between the two classes.The weight of the edge is expressed by the similarity value between the class documents,thereby constructing the undirected weighted graph of the single system.Then,based on the weighted GN algorithm discovered by the community,the undirected weighted graph of the single system is split into several connected subgraphs to obtain microservice candidates.On this basis,the microservice interface description specification is defined,and the microservice interface description method is given.Finally,the paper uses the modular quality(MQ)indicator that integrates the two aspects of cohesion and coupling to evaluate the proposed method,and uses three typical software systems to experimentally verify the method in this paper.The microservice candidates extracted by the method in this paper are compared and evaluated with the microservice candidates obtained by the methods proposed in other documents,and the results prove that the method proposed in this paper has certain advantages.