Performance evaluation in the migration process from a monolithic application to microservices

Microservices are considered as a software architecture that allows the decomposition of a system, its components or its functionalities into a set of small services, which are implemented, deployed and managed independently. In this study, the models that allow migrating a Monolith to Microservices such as NGINX and IBM are analyzed. From these models, activities that allow such migration are carefully selected and identified. In order to implement and evaluate the activities proposed in those models, an application that initially does not have any structure at the design and coding level (using PHP programming language) is applied. Then, the application's coding language changes to Java and the classes and libraries are distributed into packages. Subsequently, as it is suggested in the models, services are identified and implemented using RESTful Web Services to finally implement the microservices using technologies such as Spring Boot, Eureka, and Zuul. In the migration process, the application under study is modified at the code and design level, including patterns such as Singleton, Facade, Strangler, Single Service per Host, Service Discovery, and API Gateway, which are used to evaluate performance as a quality attribute in each migration phase. In order to obtain the performance related metrics and to analyze the advantages and disadvantages of each migration phase, Apache JMeter as tool is used. This tool is set up to generate results regarding the use of resources such as CPU, memory, network, and database access. Finally, the results show scenarios of several concurrent users who access to consult records in the database that uses the aforementioned application in each migration phase.