A DataSource is a poor concept in its existing form, as if it was intended for abstract interactions, it should not return any SQL connection that you need to interact with in the first place. Itโs also overkill, and XA is a fantastic concept that some people pay dearly for lack of reliable implementation in the world (I mean that all commercial implementers of the enterprise simply fail and open up the business ... someone suffered from for this in finance, but I will not mention the names). In general, regardless of whether it recommends Sun or Oracle, this leads to over-development and some technical error in the code (taking into account contexts, additional steps to obtain data, some external configurations ... and, in the end, the supplierโs implementation in any case). Some developed coroporate solutions that deal with consolidation, reconnecting etc are much better based on simple connections and DriverManager than the DataSource provided by DBMS vendors.
For the record, I worked with both, and I'm based on facts found in different places. And if in doubt, ask why you can see a simple JDBC URL in your Hibernate configuration throughout your business. Many just dump the heavyweigtt J2EE idea and become light ... also using simple DriverManager-based connections.
You want to get your career on the web, then go to XA DataSources and repair the failed transactions, where the poor implementation of XA is very successful.
anonymous
source share