I've been trying to read up about torque vs hibernate, but not much have come up so far...

What i'm trying to figure out, is if i should use one or the other db layer for an online community that im creating.

The framework will be Turbine, buildt with Velocity.

The scale of the things will be from around 1000-5000 online users, and 500.000+ registrered members.

I've heard some good things about both of them, but i'd like a little more information before i set it in stone and it is too late.

just think about power, performance, bugs, doc, community

I can see that there probably is more active people on Hibernate vs Torque. And perhaps also in documentation.

But i've seen no comparison on performance and not so much on ease of use. So far it seems like torque is really easy to get going with, but hibernate is a bit more difficult (so i've read also from different places).

The thing is then if there are any catches with this, and if the way it works hinders development later on.

There is simply no way how anyone can tell you what might be a problem in your project and for you in the future.

I'm pretty much aware of this....

My hopes were that some friendly soul would point out the major cons and pros of hibernate vs torque.

I've used both. From a development user perspective, the teams seem happier with the control Hibernate gives them rather than the lack of control Torque has.

Torque's easy to get running with, but difficult to work with over the long haul, it is somewhat rigid in approach to abstraction of the data access layer and hides a lot of what it is really doing from the developers. This equates to difficulties in debug or performance analysis at later points in the projects. There is little to no user or development support community, which hurts at times.

Hibernate is much "slimmer" and "open", allowing your teams to add the bits and pieces it needs, and to design a data access layer in any manner you so choose. As it is more "open" it also enables your developers to hamstring themselves easier too, but assuming you've got a good team it's manageable albeit a bit harder to get rolling with. User community and development support for Hibernate is good.

Never could get a cache manager or JTA hooked into Torque either, but this may have changed (been a year since using Torque), Hibernate gives you these out of the box.

Thanks a lot! That (with out all the details) was also almost what i expected. I also just talked with one of the turbine developers, and he suggested i used Hibernate, so i guess i'll just have to dig in and get dirty :)

Appriciate the answer.
/Martin

The Spring Framework also provides very comprehensive Hibernate support classes, which can save you a lot of coding effort.

I saw something about that yes, but how does it blend into Turbine?

i think the best way is to write a small application with 10 object and some thousand inserts an compare the time for query,insert,delete,update and make the result public.
the development of the application cost 1 day per method and to make it public cost not so much time.

For example I compare my self stupid written o/r tool and change it later to hibernate and hibernate is 20% faster on querys or 80% on updates. but before you ask, why i don't publish this stuff, it makes no sense.

but I hope my idea helps you,

gert

This is a useless test and it has been discussed many times why. There is even a thread about it right now.

Yes, of course. I don't know how many times we need to turn blue in the face saying this: performance of ORM data access code depends almost entirely upon avoiding database roundtrips via smart association fetching strategies and caching. A further determining factor of performance is the concurrency (locking) model used in the middle tier and on the database.

Performance does not depend upon the (very small) overhead added to trivial queries. No data-access intensive application on earth is limited by the speed of processing JDBC ResultSets. They are limited by such things as remote calls, disk access, locking, etc.