The docs and site talk about predictable performance, but there are no published numbers anyone can check.
Add a benchmark suite: JMH with Testcontainers, comparing Storm against plain JDBC, Hibernate, jOOQ, Exposed and Jimmer on realistic workloads, PostgreSQL first. Candidate workloads: single row by primary key, join with object mapping, projection queries, batch insert, and an N+1-prone object graph. Same schema and data for every library. Results published per release with the methodology written up, so anyone can rerun the suite and verify.
A separate storm-benchmarks repository is probably the right home; it keeps five ORMs' worth of dependencies out of the framework build.
Tasks:
The docs and site talk about predictable performance, but there are no published numbers anyone can check.
Add a benchmark suite: JMH with Testcontainers, comparing Storm against plain JDBC, Hibernate, jOOQ, Exposed and Jimmer on realistic workloads, PostgreSQL first. Candidate workloads: single row by primary key, join with object mapping, projection queries, batch insert, and an N+1-prone object graph. Same schema and data for every library. Results published per release with the methodology written up, so anyone can rerun the suite and verify.
A separate storm-benchmarks repository is probably the right home; it keeps five ORMs' worth of dependencies out of the framework build.
Tasks: