Evaluación del rendimiento entre API s (REST, GraphQL y gRPC)

Abstract

Abstract: Due to the different existing API models, when developing applications, it is necessary to choose the best option. Thus, in this work, the performance of REST, GraphQL and gRPC APIs was evaluated. For this purpose, their theoretical basis was defined; subsequently, a solution was designed and programmed for each API, using C# (.net 6) and SQL Server as database, the performance was evaluated through response time, byte transfer rate, latency and connection time, executing CRUD methods under scenarios of 10, 100, 1000 and 10000 simultaneous users; finally, good practices were documented to improve the performance of the APIs. Good practices include the use of http methods with actions and not using a verb for controllers. The REST API is the best option in many scenarios because it contemplates standardized protocols such as http; gRPC is the best option in systems with a large number of simultaneous users because it makes use of new technologies between the client and server connection, offering better connection speeds and facilitating the sending of metadata.