What's the difference between Tomcat JNDI based and Spring/Hibernate DS based connection pooling

Question

I have been looking into connection pool options and it is somewhat unclear to me what the differences in Tomcat JNDI connection pool approach is, compared to the Spring/Hibernate solution to the same.

Whilst it's possible to achieve the pooling using either 1 , 2 , the specific application we have would lend itself better to us using Tomcat given the constraints we have.

Reading about, there is some suggestion to just stick with Spring/Hibernate .

Are there any notable differences worth mentioning between each approach? What are other's personal experience of one or the other (or both) - I have successfully been using Spring/Hibernate for years now.

Answer 1

The two approaches are complementary, not mutually exclusive. In production systems, the likes of Spring/Hibernate will obtain a reference to the connection pool from the appserver, in the form of a javax.sql.DataSource , usually by looking for it on the JNDI tree. It generally considered to be the appserver's "job" to manage the connection pool and its connections.

Remember, JNDI is just a place for registering objects for sharing, it does in itself mandate any given connection pool mechanism. The app server creates and configures the pool, and the applications (via Spring/Hibernate/whatever) use it.

It's just as valid, however, for the applications to configure and manage the connection pool themselves. This does mean a bit more work for the application, though, with less reliance on the appserver.