Philosophy of transaction management in a job

[vijaynairis]

Hi,

I have got a simple job which reads from a table in a database and does some processing on it (CustomWriter) and persists to another table in the database

I use the following principles for transaction management

JDBCCursor I presume opens a new transaction irrespective of the incoming transaction attribute on the thread local....I then call a service class method which has a @Transactional attribute...which runs in the transactional context of the original transaction opened as per the Batch context....

Am i right in this philosophy? I see no difference in performance irrespective of using Batch or without Batch because I am pretty sure I am missing something out here to speeden up my process using Batch...

is there any facility in 2.1 Batch which will help me speeden up the same ?

Any inputs would be highly appreciated

TIA...VJ

[Marten Deinum]

Post some code and configuration.

I assume you have set a commit interval, you aren't messing around with transactions yourself, you have transactional tables.