Study of 362 open-source Java projects reveals DBMS adoption and polyglot persistence trends

A team of researchers from multiple universities conducted a historical study of Database Management System (DBMS) adoption in 362 popular open-source Java projects hosted on GitHub. They analyzed source-code heuristics to track usage of top DBMSs ranked by DB-Engines, covering both relational and non-relational systems. Key findings show MySQL and PostgreSQL are the most popular relational DBMSs, while Redis and MongoDB are the most used non-relational systems and tend to remain stable once adopted. In contrast, HyperSQL is more frequently replaced as projects evolve. The study also documented widespread co-use of multiple DBMSs—a pattern known as polyglot persistence—where projects combine systems like MySQL with Redis to handle different data needs. Additionally, Object-Relational Mappers (ORMs) are commonly employed to mediate application-DBMS interactions.

These findings have practical implications for developers and architects. The stability of Redis and MongoDB suggests they are fit for long-term use, while the frequent replacement of HyperSQL indicates it may be a temporary or transitional choice. The prevalence of polyglot persistence underscores that modern applications often require a mix of specialized databases rather than a one-size-fits-all solution. For educators and vendors, the data provides empirical insights into real-world usage patterns, helping guide curriculum design and product roadmaps. Overall, this study offers data-driven guidance on DBMS selection, migration, and co-use in open-source software ecosystems.