hi!

i'm using hibernate (JDK 1.4.2) with sapdb (7.4 the latest stable)

when i want to use hibernate for an import program then hand coded sql seems to be more than twice(!) faster than hibernate.

each object to insert is just a plain object with three attributes:
long (as id)
String
Timestamp

here the code sniplet for the insert:

for (each object to insert) {
Session sess = sessions.openSession();
Transaction tx = sess.beginTransaction();
sess.save(bosEx);
tx.commit();
sess.flush();
sess.close();
}

"bulk" inserts are not acceptable because i shouldn't :-) loose data!

any hints or tips?
many thanx, blip

First of all, I would move the session open/close and the transaction outside the loop. These tend to be the most expensive operations - you don't want to repeat them for every single insert.

Also, if it needs to work with very large amounts of data, you may want to use jdbc directly.

Joe

using the same session results in an OutOfMemoryError .....
i need to know if exactly one entity was correctly inserted.

SORRY - my mistake!

i didn't activate the PreparedStatement Pooling *uups*
now i'm "just a little bit (20%) faster with handcoded sql (a commit for every object)

BUT with a "bulk" insert (commit after every 1000 objects) HIBERNATE is 8% faster ......

how to enable PreparedStatement Pooling?

try using session.clear() after the flush.