(further lies, damned lies, and benchmarks)

Hi all,

I would like to direct you to the following benchmark that covers Hibernate:
http://www.polepos.org

The published results use a 2.x version of Hibernate and the tests and the code do not look like they are ideal yet.

I just read the Hibernate Performance Q+A and Gavin's rants on this page:
http://www.hibernate.org/157.html

I fully agree with what Gavin writes: Simple benchmarks usually are meaningless if they do not uncover what is run in memory and what is written to disk.

However I do really like the solution of the PolePosition benchmark to run number rows and to plot curves to get an idea on scalability. I also think the overall framework is convenient enough to add further tests easily.

My suggestion:
Why don't you embrace the PolePosition benchmark and contribute own ideas? Maybe it can develop into a strong standard benchmarking platform, so you will never have to evaluate another benchmark again, Gavin? Could the code behind the perftest target maybe be ported to Poleposition?

I do like one aspect about benchmarks:
If you spend weeks and months optimizing performance you have something to show users and management.

Best,
Carl

You really think that the reason people are using RDBMS over OODBMS is performance?

No, I think people are using RDBMS because it's either their installed base or because they do not know that ODBMS exist.

If people need performance to store real OO structures (trees and inheritance hierarchies), ODBMS are the natural choice because they are optimized to do it: faster and easier to use.

The PolePosition benchmark shows this quite clearly, both in the performance graph and in the source code.

Carl, this is a warning. Do your marketing somewhere else.

Christian,

I am trying to contribute an idea:

The PolePosition benchmark would be a chance for you to point out the performance strengths of Hibernate in comparison to other O-R mappers. I am pretty sure that Hibernate is the best O-R mapper available and it will benefit from publicity that mentions it.

If you think that's a bad idea, please tell me why.

What are you afraid of? db4o and Hibernate do not compete at all. Look at the installed base out there. People are investing somewhat USD 20 billion into relational databases. They are are all potential Hibernate users and no customers for us.

Instead of seeing us as a competitor you should see us as an ally for object persistence.

You are very welcome to post to our newsgroup, if you post positive content, that people can benefit from. No warnings.

Best,
Carl

Carl, nobody gets upset about competition, people get upset over FUD benchmarks.

Take it offline, guys.