Spring Data JDBC performance

Hi

We have a Spring Data JPA (Hibernate) project that we would like to migrate to Spring Data JDBC. The main reasons are simplified configuration and model mapping. And we thought that it would lead to better performance.

However we did some benchmarks (with default settings) using JMH (H2 database) and it turned out that in most cases the performance has decreased. For example, we have the following repository and query method:

public interface ProductRepository 
	extends CrudRepository<Product, Integer> {
	
	Product findByName(String name);

And the benchmarks showed the following execution time (in ns): Spring Data JDBC - 80899 Spring Data JPA - 14124

So is it predictable? Or we missed something in our configuration/tests?

[petromir]

It seems, we also do not properly cache the results of constructing the SQL statement from the method name, resulting in some significant overhead. You can workaround that by providing an explicit query. The missing caching is something we’ll fix.

@schauder Is this something to expect in the upcoming version 3.0?

[schauder]

Was it done intentionally? Yes and no. I intentionally added the annotation to demonstrate the effect of not caching the generated query. Changing the query semantics was a mistake on my side.