Study of Graph Database Implementation on Dynamic Table of BPS Website

BPS is a government agency that provides statistical data. The process of disseminating BPS statistical data to the public is carried out through the BPS website in the dynamic table feature. This feature requires high database performance to handle large data volumes. Because of this, this research implements one type of non-relational database, Graph database, then compares its execution time with the current system (relational database). The implementation of graph database using Neo4j resulted in 3,365,778 nodes and 16,810,408 relations. The performance measurement results show that Neo4j can be 1.19 to 30.26 times faster than the current system under the condition of large data and large number of users. However, PostgreSQL is also 2.74 to 7.87 times faster under the condition of small data and small number of users. This shows that Neo4j is superior to PostgreSQL on read operations when the amount of data gets larger, the number of users gets larger, and the query complexity gets higher.