Does the use of JCA (locally, on the same box as tamino, not in a distributed environment) circumvent the X-Machine commands with native Java connections of some sort? Or does JCA use HTTP and hence not improve connection performance? I like the platform-agnosticism of X-Machine, but HTTP adds uneeded overhead to a lot of pure-java apps we are developing (e.g. an HTTP request to an app on Tomcat becomes an HTTP request to Apache through the java API- it’s kind of ridiculous). Unfortunately, though I’m an old servlet hand, I’ve never grasped the deal with the whole J2EE stack, and don’t know how Tamino, the JCA and whatever (app server? EJB in a container?) integrate, and what advantages it might provide (beyond transactions).

Also, what about JCA in a distributed environment? Does JCA use RMI to communicate with Tamino instead of HTTP? If so, does RMI beat HTTP speed-wise? Scalability-wise?

Thanks in advance for any help.

JCA does not communicate with the X-Machine natively but still uses HTTP for its communication. JCA is all about providing connectivity to non JDBC databases within the context of a J2EE server. As part of the JCA implementation there is a Tamino Resource Adapter which provides the connectivity to the Tamino database. As I’ve mentioned this is HTTP and does not use RMI.

If you are developing servlets then you need to use the standard Tamino API for Java. The JCA is required when developing Enterprise Java Beans that want to persist their data to Tamino rather than a JDBC compliant SQL database.

Hope this helps.

It does indeed help. Thanks for the info.