Hi Alan,
Thanks for the quick reply. Yes, a new Oracle instance is created for every request. I created a wrapper class (SparqlConnection) for the oracle.spatial.rdf.client.jena.Oracle object. The wrapper contains an attribute of type Oracle.

1) my constructor contains this code:
oracle = new Oracle(jdbcUrl, user, password);

2) SparqlConnection executes the query

3) When the query is done call close() method. The close() code looks like:

try {
if (oracle != null) {
oracle.dispose();
}
} catch (Exception e) {
System.out.println("Error closing Oracle connection. Error: " + e.getMessage());
e.printStackTrace(System.out);
} finally {
oracle = null;
}


Is there a connection pooling class or do I need to create my own? When you say "we add a shutdown hook so that unfinished transaction can be rolled back when VM shuts down" are these hooks only added when the executed statement is an INSERT/UPDATE/DELETE? Or is the hook added for SELECT statements also?

Thanks,
Chuck

I will create a connection pool for my Oracle connections. That should help. However, I have two other questions:

1. Is a hook created for the SELECT statements? If so, is this really necessary? Does a SELECT statement need to be rolled back when the JVM shutsdown?
2. If the hook is created for each Oracle connection, does calling commitTransaction() remove the hook?

Thanks,
Chuck