Hibernate ORM Mapping and Runtime Behavior

Hibernate is a state machine with a SQL generator attached. Most production surprises come from ignoring the state machine, not from a missing annotation. Decide mapping deliberately, keep the persistence context small, and let the schema live in migrations.

When to use

• Choosing between @EmbeddedId and a surrogate key, or between @ElementCollection and a JSONB column.
• Fixing LazyInitializationException, MultipleBagFetchException, or a cascade that writes rows nobody expected.
• Setting up batch inserts, ordered updates, and the identifier strategy that unblocks batching.
• Proving or eliminating an N+1 with Hibernate statistics, understanding the four entity states and dirty checking cost, and deciding whether a StatelessSession or the caches help.

When not to use

• SQL dialect, index selection, or plan analysis: hand off to spring-boot-postgresql.
• Repository interfaces, Specifications, or pagination API design: hand off to spring-boot-data-jpa.
• Migration files, baselines, or schema history: hand off to spring-boot-flyway.
• End-to-end latency, caching layers, and throughput remediation: hand off to spring-boot-query-optimization.

Ownership and sibling boundaries