HikariCP là gì? Tại sao là default connection pool trong Spring Boot?

HikariCP là JDBC connection pool library — quản lý pool kết nối DB thay vì mở/đóng connection mỗi request.

Tại sao HikariCP là default (Spring Boot 2+):
- Nhanh nhất trong benchmark (TomcatCP, DBCP2, C3P0) — minimal overhead per borrow/return.
- Lightweight (~170KB JAR).
- Reliable — được test kỹ, ít bug.

spring:
  datasource:
    url: jdbc:postgresql://localhost/mydb
    username: postgres
    password: secret
    hikari:
      pool-name: MyPool
      maximum-pool-size: 20       # max connections
      minimum-idle: 5             # min idle connections
      connection-timeout: 30000   # ms chờ lấy connection
      idle-timeout: 600000        # ms trước khi idle connection bị đóng
      max-lifetime: 1800000       # ms tối đa connection sống
      leak-detection-threshold: 60000  # ms — cảnh báo nếu connection bị giữ quá lâu

Pool size best practice:
- Không phải càng nhiều càng tốt — DB có giới hạn concurrent connection.
- Formula: pool_size = (core_count * 2) + effective_spindle_count (HikariCP docs).
- Postgres recommend: 10-20 connections per app instance.

Leak detection: set leakDetectionThreshold để phát hiện connection không được trả về pool (code bug).

HikariCP is a JDBC connection pool library — manages a pool of DB connections instead of opening/closing one per request.

Why HikariCP is the default (Spring Boot 2+):
- Fastest in benchmarks (vs TomcatCP, DBCP2, C3P0) — minimal overhead per borrow/return.
- Lightweight (~170 KB JAR).
- Reliable — well-tested, few bugs.

spring:
  datasource:
    url: jdbc:postgresql://localhost/mydb
    username: postgres
    password: secret
    hikari:
      pool-name: MyPool
      maximum-pool-size: 20       # max connections
      minimum-idle: 5             # min idle connections
      connection-timeout: 30000   # ms to wait for a connection
      idle-timeout: 600000        # ms before an idle connection is closed
      max-lifetime: 1800000       # max ms a connection stays alive
      leak-detection-threshold: 60000  # ms — warn if connection held too long

Pool size best practice:
- More is not always better — the DB has a limited concurrent connection count.
- Formula: pool_size = (core_count * 2) + effective_spindle_count (HikariCP docs).
- Postgres recommends: 10-20 connections per app instance.

Leak detection: set leakDetectionThreshold to detect connections not returned to the pool (code bugs).