Efficiency study of GraphQL and REST Microservices in Docker containers: A computational experiment

Introduction: In the constant evolution of technology, implementing new services in computer systems is crucial. However, the integration of these services presents problems and certain challenges in the deployment of applications. Technologies such as Docker and microservices architectures are alternatives to alleviate such integration. The aim was to compare the performance efficiency between microservices architectures implemented with GraphQL and REST, deployed in Docker and localhost environments. Methods: A computational experiment was conducted following the Wholin methodology to compare the performance efficiency of microservices architectures. The experimental design consisted of deploying both a GraphQL API and a REST API with identical functionalities in Docker containers and a localhost environment. Both APIs were consumed under controlled complexity and data volume conditions, ensuring a fair evaluation. Results: The experiment showed that the average response time in the Docker environment was significantly lower compared to the localhost environment. Also, the GraphQL API outperformed the REST API. In addition, a research artifact including all the study materials was published on Zenodo to support the replicability of the experiment. Conclusion: The architecture deployed in Docker is more efficient for microservices execution, particularly when GraphQL is used.