Oracle Database Discussions

The way the ORMs are designed, integration with application servers are relatively simple, and all provides the same features. Also since WebLogic have been around for a while, all ORMs are all well tested in this configuration.

Hibernate has lot more users, and is likely very often used with Oracle DB, so you can expect not much bug against Oracle DB, maybe even less bug than EclipseLink, which is not much used. EclipseLink does provide support for some esoteric Oracle DB features like hierarchical and flashback queries.

OpenJPA and DataNucleus are also JPA compliant. It’s likely that Open JPA has a higher user base than EclipseLink, so less unknown bugs.

Oracle paying support is well known to be a bad joke. It’s a negative return to use this channel, even if they would be free. So in reality, you end up to use the open (free) forum to get support.

What’s was lacking with Hibernate before is Dynamic Fetch Planning, but they now have some support, see http://opensource.atlassian.com/projects/hibernate/browse/HHH-3414. OpenJPA was the first to implement this must have.

EclipseLink has query in memory, which can be used, but the API do not help to leverage it, and EclipseLink’s leadership made it clear that they are not going to make it better, instead they want to push Coherence cache.

Hibernate has an open API for second level cache, which mean you can get out of problem by using another implementation, for example, EHCache seems to be professionally tested, so I would be surprise you find obvious bugs.

I cannot comment on Hibernate source code quality, but I can tell you that locking mechanism in EclipseLink is used to be very fragile, and many concepts are dispersed over the code base.

The runtime monitoring of Hibernate have always been great due to the fact that JBoss have always been strong on JMX, EclipseLink has not much usable features on this.

If I would be you, I would consider OpenJPA or Hibernate instead of EclipseLink, the main reason is that because EclipseLink has a so low user base, I have found lot of obvious bugs in production, like if I was the only user of it. Then, when I submitted bugs to the small development team, which do not encourage user base contribution, they were too busy trying to keep up adding the JPA interfaces on top of their existing proprietary APIs.