A common mistake is not tuning the init and shutdown ordering of your connection pools versus the transaction kernel - so please make sure you avoid this pitfall:

Your REST service is invoked and left pending. How can you rollback?

So you want elastic scaling. What about transaction recovery?

Our customers regularly ask for disaster recovery options in combination with our JTA/XA implementation.

This post was based on old content and is no longer relevant.

Visit our Distributed Transactions Tutorial for equivalent content.