PostgreSQL for Spring Boot

Treat PostgreSQL as a system with its own cost model, not as a store for Java objects. Type, index, isolation, and pool choices are correctness and latency decisions the application inherits whether it asked for them or not.

When to use

• Choosing column types, constraints, or index types for mapped entities.
• Reading EXPLAIN (ANALYZE, BUFFERS) output and deciding why a plan is wrong.
• Diagnosing deadlocks, lock waits, bloat, stale statistics, or long-running transactions.
• Choosing an isolation level, a FOR UPDATE strategy, or a SKIP LOCKED work queue.
• Configuring HikariCP and pgjdbc: pool size, timeouts, leak detection, statement timeouts.
• Writing native SQL with CTEs, window functions, jsonb operators, or set-based DML.

When not to use

• Entity mappings, associations, fetch strategies, or batching: hand off to spring-boot-hibernate.
• Repository interfaces, Specifications, or pagination API design: hand off to spring-boot-data-jpa.
• End-to-end slow-request work across cache, network, and application CPU: hand off to spring-boot-query-optimization.
• Versioned migration files, baselines, checksums, or repair: hand off to spring-boot-flyway.