Editor's note (no single primary source): "N+1 queries and open-session-in-view; a badly sized connection pool; blocking calls with no timeouts; too much logging at DEBUG; pulling in starters I don't need, which slows startup; and no caching or measuring before tuning." Source for parts: API latency optimization (N+1, HikariCP pool sizing, measuring), SQL databases
Sources (0)
See inline citations in the answer above.