I queried for information regarding performance differences regarding the table per subClass (with discriminator) vs joined subClass, but did not find the information to help me. I am pretty sure that the table per subClass (with discriminator) is the better performer. Does anybody have any tests or experience to confirm my suspision.

Thanks.


sorry, previously posted in wrong forum.

Actual perfomance is depends on many conditions: DB type, size of tables, presence of indexes, selectivites of descriminator column, type of queries and many others conditions....

I agree, I guess I was thinking that a read on the database with the discriminator would be more efficient because the superClass would know the subClass and would create a better join clause that what would be created with the joined subClass.

Not nessary. For example, if discriminator column have poor selectivity (many for one value, but seldom for another), then selectings for exact subclass may result bad perfomance due bad query plan generated. This is especially noticiable with Table per class hierarchy strategy.

I would recommend you to build prototype model for your data ant test for speed your queries. Because your data and your queries may significantly differs from ours. :-)