Hikari Connection Pooling with Spring Boot
summary
Hikari connection pooling for Spring Boot shows concrete knobs and measurable effects: min=5 max=10 scales to demand and uses a housekeeper every 30s to retire idle connections; connection timeout=5s caps response time and produces timeout exceptions when pool is saturated. Experiments use JMeter with Grafana/OpenTelemetry/Prometheus. Network latency needs more connections (bedroom vs EC2: ~6 vs 2), slow queries hog connections (1800 cheap : 50 slow), separate datasources rescued latency, and reducing connections (248→96) cut median latency 100ms→3ms.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:09Connection pool basics
- 3:21Hikari pool features
- 6:12Configuring connection limits
- 8:46Idle connection management
- 10:57Connection timeout handling
- 13:06Mass extinction events
- 15:19Maximum lifetime parameter
- 17:25Keep-alive mechanism
- 17:45Common performance pitfalls
- 20:12Query execution imbalance
- 23:05Cluster database issues
- 25:57Database resource limits
- 27:02Real-world scalability test
- 28:50Connection limits impact
- 29:10Q&A and tools
related talks
50mREX: HikariCP Optimizations in Java/Spring Boot/OpenShift
44mPerformance oriented Spring Data JPA & Hibernate
45mWhen ORM Becomes OMG: Performance Pitfalls in JPA and Friends
49mPerformance oriented Spring Data JPA & Hibernate
55mBootiful Spring Boot 3
47mWhen ORM Becomes OMG: Performance Pitfalls in JPA and Friends
30mBuilding Scalable Spring Applications with YugabyteDB
49mPanta rhei: runtime configuration updates with Spring Boot