Okay, that's at least kind of good news. Well we are using Spring for our Data-Access Layer when it comes to not using Hibernate directly. I don't really know where to configure the second level cache correctly there in our application tbh.

But I don't really like that I have to manually make sure that no caching is used. And just to show you in numbers: For 1 result list we have to load approx 50-60 entities from cache. And the hydratation/lookup process seems quite a big deal tbh. I will have to wait on the benchmarks for this as I have to get done and since it will be a time since my proposed feature will be implemented (at least I think that) I think I will have to live with my hack.