Interconnect is currently in the frame as the Hub and Spoke for a solution integrating various vendor components.
However one item I would like to get clarity on; the concept that for a message to be propagated via Interconnect a commit must be issued. e.g. Doing data entry against Oracle Database; the inputted data will be validated against details from another database and/or a call to a rules engine using Interconnect. To get the message out to Interconnect my current understanding is that a commit must be issued in the Data Entry Database ==> commits the entered data to the database.
However if the validation fails (e.g. person id not valid ) after the call to the other database / rules engine , concerned we may have previously committed data in the Data Entry Database which is not correct. Ideally would like to issue a commit on the data entry database only when the inputted data has been successfully validated.
Is a commit required to propagate messages via Interconnect ?
If so is it possible to restrict the scope of the commit to Interconnect Message call by using the Autonomous Transaction construct ?
Are there other approaches ?
Have searched various places on Oracle web-site but can't find a definite answer to the above questions, so all advice most welcome.
Any pointers to what I should read up on which would aid my understanding on Interconnect much appreciated.