Migration of Monolithic to Microservices With an Extraction Design Pattern in Single Sign on (SSO) Module Using Graph Neural Network (GNN)

In software system development, Mololithic is an architectural model that is built in the same codebase. For example, on an online shopping site, there are several feature sections such as user authorization, shopping cart, payment, shipping, and others, all of these features are built in an integrated manner on the same codebase. This is different from the microservice architecture approach. In microservices, each feature or component will be an architecture that is separate from one another. Each of these services will later communicate with each other via a special protocol. Mololithic is increasingly being abandoned because it does not meet scalability needs, if there is a surge in users the Mololithic model tends to be unstable.As happens in the SSO module, in the SSO module process itself there are several functions including registration, authentication, user management, back end API and several other functions where these functions are combined in one unified module. Therefore, a Microservices Architecture model is needed to handle the needs due to the surge in users. Scalability can be regulated because the codebase separation process depends on needs. The migration process from monolithic to microservices is not easy, it requires sufficient knowledge regarding the application to determine the number of nodes and determine the number of services. This research will separate the codebase to determine the number of services using extraction design patterns on function entities, and system logs used in the SSO module at one institution in Indonesia.It is hoped that by extracting functions from the source code and data transaction we will get how many services are running and the nodes that will be used. The modeling process regarding the number of nodes and services will be carried out using a Graph Neural Network (GNN)