Optimizing MyBatis-Plus Batch Insert Performance with rewriteBatchedStatements and Multithreading

When synchronizing tens of thousands of base records between systems, using MyBatis-Plus's default saveBatch() can take many seconds because the JDBC URL lacks the rewriteBatchedStatements=true property.

Adding the property to the connection string, e.g. jdbc:mysql://host/db?useUnicode=true&characterEncoding=UTF8&allowMultiQueries=true&rewriteBatchedStatements=true , reduces the insertion time of 10,000 rows to a few hundred milliseconds.

The article then demonstrates the default saveBatch() implementation, showing the source code of the method and a unit test that inserts 200,000 rows, which takes about 10 seconds.

To achieve better performance, a custom RootMapper interface extending BaseMapper is defined with insertBatch and updateBatch methods, followed by custom injector classes InsertBatchColumn and UpdateBatchColumn that generate the appropriate SQL for bulk operations.

The custom injector MysqlInjector registers these methods, and a configuration class exposes it as a Spring bean.

Using the custom batch insert with a batch size of 5,000 rows reduces the insertion time to 4–5 seconds, while the same data inserted with the default saveBatch() (with the JDBC property enabled) also achieves similar speed.

Further performance gains are obtained by combining saveBatch() with multithreading: a fixed thread pool of five threads processes the data in parallel, bringing the total time down to about 3 seconds.

Finally, the article shows a multithreaded approach using the custom insertBatch method, which completes the 200,000‑row insertion in roughly 2 seconds, and compares all results, concluding that enabling rewriteBatchedStatements and applying multithreading provide the most efficient solution.

Overall, the guide provides practical code snippets, performance measurements, and recommendations for developers needing fast bulk data operations in Java backend projects.