Database replication is a critical technology that needs to be implemented to improve database performance for applications that require fast data access and high reliability with many users. Very large data query process requests by applications impact decreasing database server performance because many applications use only one database resource. One solution to this is replication, placing the database server and several replications in different locations, which allows database performance to be maximized. This requires a study of research related to replication strategies, especially the implementation of multi-location database geo-replication mechanisms, which aim to ensure the availability of services and replication processes run well and that consistency between database servers can be achieved. The increasing number of replicated databases with different geographic locations will make data access load requests from applications can be directed to several replication servers and reduce the delay time to user responses (low latency). This research implemented a geo-replication database for three database servers on the Ubuntu 22.04.4 LTS Virtual Private Server located in Singapore, West Jakarta, and South Jakarta, using the MariaDB database version 10.6.18. To measure the replication results related to time, consistency, and accessibility, a database replication query processing time test was carried out using queries with data from 250 to 150,000 records. The test results from the consistency side can be seen from the same number of data records on the master and slave replication databases, and from the accessibility side, data query requests can be served from three database server locations.