The article references experiment details
https://github.com/PVRLabs/experiments/tree/main/spring-boot..., but I hadn't explored it much.
What surprised you? I thought the experiment simply demonstrated that there is a noticeable overhead in using JPA with small apps.