http://www.middlewareresearch.com/torpedo/results.jsp

I am not sure what the involvement of the Hibernate team will be with this effort. Just want to bring this to everyones attention

mapping file and query were not tunned .... not a really serious test guys

too all who want to avoid these kind of results, 2 words : lazy and fetch query keyword

Oh, and of course: the power of an ORM tool is not "only" the number of db access.

Anyway, real serious tests should be done soon.

After one hour of standard optimization (they didn't do any, not even lazy loading or eager fetching), Gavin got the number of queries in this "benchmark" down to 6 and sent them the optimized mappings and results. Yes, thats six, down from sixhundred. This should give you a hint about the quality of their "benchmark". I would recommend to simply ignore it.

In fact, I already have a much better "number of SQL statements benchmark" ready for CaveatEmptor, and we will make it public soon. It will be easy to implement it in other persistence layers, so you can have a try and see if you can beat Hibernate. More next week...

Basically these guys mapped all classes and associations lazy="false", so they retrieve the entire database in every transaction.

Hilarious.

Does anyone take this company seriously? They've demonstrated in the past (remember the .NET vs. J2EE fiasco) that they are anything but honest. TheMiddlewareCompany (TMC) is one of those "research" companies that takes money from vendors in order to find the results that vendor wants. Vendors have to pay them for "briefings" - which is code for writing fluff pieces on the vendor's latest "breakthrough". TMC doesn't do research, they do public relations.