Analisis Performa Dan Kompleksitas Application Programming Interface Pada Arsitektur Monolithic Dan Microservice (Studi Kasus: Gocommerce)

Abstract

As applications become large and teams grow, monolithic architectures have several disadvantages such as applications become difficult to understand and modify, continuous deployment is difficult, scaling applications becomes more difficult, and requires long-term commitment to the technology stack. Research was conducted to determine the performance differences between APIs built using monolithic and microservice architectures in tests with certain device specifications and measure the level of complexity. This research uses GoCommerce public application source code obtained from the GitHub repository. The modules that will be tested in this research include user, report and order modules. Data from each module will be created using dummy data with a total of 5000 rows of data for performance testing. Performance testing is done with Apache JMeter tools, while complexity testing is done with CK-Metrics. APIs built with monolithic architecture get better performance on low server specifications (1 Core 1 GB RAM) with an average throughput value of 39.83/sec and an average percentage error of 22.87% compared to microservice architecture which gets an average throughput value of 0.66/sec and an average percentage error of 73.16%. While APIs built with microservice architecture tend to be better at higher server specifications (2 Core 2 GB RAM, 4 Core 4 GB RAM, and 6 Core 16 GB RAM) with an average throughput value of 13.97/sec and an average percentage error of 0.40% compared to monolithic architecture which gets an average throughput value of 16.57/sec and an average percentage error of 16.47%. API with monolithic architecture has lower complexity than microservice architecture with the calculation results of WMC value 5 compared to 11, RFC value 130 compared to 131, LCOM value 1 compared to 4, DIT value 3 compared to 6, and NOC value 3 compared to 6.