I have to perform concurrent inserts in UDB 7.2 on Windows 2000.
The current configuration sets the lock size to "Row".
I understand that a process performing an insert has an exclusive lock on the next key. This prevents another process from accessing the table to insert another row until the first one commits.
In Sybase ASE for example, this can be circumvented with table partitioning (which does not require multiple nodes as opposed to UDB 7.2) where the lock is acquired on a page. There are good chances the second process will request the lock on another page.
What is the equivalent in UDB 7.2?
Any help is welcome.
P.S. To give more information on the context, I have a bunch of applications that run in parallel. They read a message from an MQSeries queue, insert it into a table, perform some processing, put them in an output queue and commit (using MQSeries as the transaction manager for 2-phase commit). These applications are written in Java.