Experimentell jämförelse av Cassandra, MongoDB och MySQL

Context. Performance and resource usage is often an important factor for database driven services. Depending on the purpose of the application, choosing the wrong type of database may require unnecessary hardware upgrades or replacements. Objectives. This study investigates the differences in performance and resource usage between different SQL- and NoSQL-databases. The study is a comparison that focuses on how the databases perform in different applications. The purpose is to give a hint of which database may be the most appropriate depending on the application. Methods. The different databases are compared using experiments, executed with the help of a testing tool developed to measure the performance and resource usage for the different databases. Results. Several tests have been executed on Cassandra, MongoDB and MySQL that are the databases used in this study. Where the performance is measured as operations per second and the resource usage is measured as the activity of CPU, RAM and harddrive. Conclusions. We conclude that the databases perform differently well dependning on the application. In some of the cases MongoDB performed the best, while MySQL performed best in others. The resource usage also differs a lot, like the performance, dependning on the application. As a conclusion it’s recommended to test what database may suit the specific situation depending on the application