IMPLEMENTATION OF POLYGLOT PERSISTENCE AS A DATABASE OF THE GANELOVE DATING APPLICATION

• Main Author: Kiranadewi, Andjani
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/74168
• Institution: Institut Teknologi Bandung

The COVID-19 pandemic forced people, including college students, to do activities at home. Because of this, social interactions were restricted to online-based interactions and overall lessened. Thus, a dating application called Ganelove was developed to overcome this problem. However, during the operational period, several issues were identified. First, the database’s bandwidth consumption was high, thus increasing operating costs. In addition, users considered that the application’s response time was not fast enough. Therefore, there is an opportunity to improve the application’s database system. The development of a database consists of five stages: identifying the application’s requirements, creating a conceptual model from the requirements, selecting the DBMS used, converting the conceptual model to a logical model, and designing the physical model. The database selection uses the application’s non-functional requirements and the relations between tables as the criteria. From the development steps, it is determined that the database system is developed using more than one DBMS, or, in other words, using polyglot persistence. The two DBMSs chosen are MongoDB and Apache Cassandra. To connect these two DBMSs, a Node.js-based API is used. The database system is tested by performing validation on the queries needed by the application and comparing its performance against the previously used document-oriented database. There are two parameters for comparison: the database system’s execution time against a certain number of records in the database and the database system’s capability to handle a certain number of connections at a time. According to test results, the database system that uses polyglot persistence performs better than the database system that only uses one kind of DBMS in terms of execution time and concurrent connection handling.