The idea is good, the implementation so-so and some details are a bit lacking.

1. You use different Docker images for each stack? Why? With Rust, you can easily use FROM scratch; with GraalVM, it's a bit more involved but you can achieve the same.

2. You mention yourself that you execute two requests for Rust, POST then GET. Why don't you use the RETURNING clause so you get the result directly from the INSERT (cf. https://www.postgresql.org/docs/current/sql-insert.html)?

3. In Spring Boot, you use Spring Data JPA that optimizes the SQL for you but craft the SQL manually in Rust. An easy optimization is to SELECT specific columns instead of *.

4. Worse, you use bind instead of persistent; the latter allows you to cache and reuse the query (cf. https://docs.rs/sqlx/latest/sqlx/query/struct.QueryAs.html)

5. If you use a database, the test is going to average the stack's results. I assume that you want a "real-world" scenario. In this case, why 8 records? There should be 800.000 records. Or 8 millions!

The above improvements should get you on the path toward a more apple-to-apple comparison.

Good luck!