Oracle Link for Distributed Transaction

Question

In my application, I've to integrate two system (or I would say two separate instances of Oracle databases) through a Java Program.
Program will copy some data (approximately 100 records/hour) from one instance of Oracle database and transfer the same to other instance of Oracle and Vice versa. Since this is going to be a stand alone batch application, I thought of using Distributed Transaction Manager like JBOSS TM for controlling distributed transaction.
Is this approach right? Or can you suggest me any other better approach for this job?

Answer 1

853627 wrote:
I do agree with your Oracle Link explanation, and Yes that would be ideal to have in this case. For some reason, I'm forced to use the Java way.
Why i'm using the distributed transaction because, I'm following certain steps to complete one iteration:
1) Read the data from Database A,
2) Move the data to Database B
3) After successful execution of Step 1 and 2, There is another table in Database A where I need to mark that certain records have been migrated from Database 'A' to 'B'
4) After that I've to update a summary table with details of the migration like : Number of records read, Time when this routine completed, any validation error in records etc..Distributed transactions require database level support and complications.
But I don't see anything that suggests you need this.
First step is no transaction based.
Only failure point is between 2 and 3/4. And I suspect you could handle that with a tracking table in A.
1) Start transaction, Read the data from Database A, Update tracking table, end transaction.
2) Move the data to Database B
3) Start Transaction
4) After successful execution of Step 1 and 2, There is another table in Database A where I need to mark that certain records have been migrated from Database 'A' to 'B'
5) (Presumably A) After that I've to update a summary table with details of the migration like : Number of records read, Time when this routine completed, any validation error in records etc..
6) Clean tracking table
7) End transaction.
If there is a failure, then use the tracking table to facilate correction.