I expected JDBC to have a smaller memory footprint, but the startup time difference is quite substantial. I noticed the experiment record linked at the bottom. Does it also include the complete source code for both application variants?
You're right, I misread the results. But by the way, it seems additional overhead also arises from building entity models in memory and initializing other heavy classes, such as EntityManagerFactory.
I expected JDBC to have a smaller memory footprint, but the startup time difference is quite substantial. I noticed the experiment record linked at the bottom. Does it also include the complete source code for both application variants?
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.
I think the startup difference comes from classloading costs. It simply takes much longer for Hibernate stuff to load initially.
You're right, I misread the results. But by the way, it seems additional overhead also arises from building entity models in memory and initializing other heavy classes, such as EntityManagerFactory.
Also, thanks for sources!