As stated before, just because microservices is the latest trend, doesn’t mean that you have to change your system architecture accordingly. Other than the organisational and technical challenges ahead, it may not necessarily even result in a better system performance for your application.
Although there are not that many researches around this area, a comparison done by Al-Debagy & Martinek (2018) shows interesting trends. The research that focuses on the performance of microservices and monolithic applications and uses JHipster to generate web applications consisting of Angular framework and Spring Boot and had three services: A register of all the components called JHipster registry, a backend service that provides API to the frontend and a microservices API gateway. The JMeter software was then used to test the performance and measure response time and throughput. Also, load testing and concurrency was done to see how the application behaved under heavy load.
(Al-Debagy & Martinek, 2018)
The conclusion was that both architectural style can behave similarly under normal load. The only scenario when microservices shine is under heavy-load and concurrency. What needs to be highlighted more is that the monolithic application can respond faster (due to lower over-head as there’s no remote call) so it provides higher throughput on average.
Rererences
Al-Debagy, O., & Martinek, P. (2018, 21-22 Nov. 2018). A Comparative Review of Microservices and Monolithic Architectures. 2018 IEEE 18th International Symposium on Computational Intelligence and Informatics (CINTI)