The benchmark that made me change my mind about Jakarta EE in 2026

A revised benchmark compared the same DB-heavy API on Spring Boot, Payara Micro, and Embedded GlassFish under stricter controls.

After fixing the methodology — uniform JDK, fixed heap, separate warmup, longer test windows, and explicit pool settings — the initial “GlassFish is fastest” result changed.

The improved benchmark showed that better testing discipline can overturn an early performance conclusion, especially when database time is properly attributed.

Spring Boot remained the best fit for new Spring-based projects, while Payara Micro suits existing Payara/Jakarta stacks and Embedded GlassFish stays useful for lightweight Jakarta executables.