Learning iBATIS - Transactions

Transactions are one of the most important concepts to understand when working with a relational database.

What is a transaction?
In the simplest terms, a transaction is a unit of work usually involving a number of steps that must succeed or fail as a group. Should any step in the transaction fail, all steps are rolled back so that the data is left in a consistent state.

Transaction can be very small and basic, perhaps consisting of only a couple of SQL statements that change data in a single table of a single database. However, transaction can also become very large and complext. A business-to-business transaction could even leave the realm of computers and require physical interaction with human beings(e.g. a signature). The topic of transactons could easily fill a book of its won, so we'll only consider four scopes of transactions that iBATIS supports:
 #1 Automatic - For simple, single statements that don't require an explicitly demarcated transaction.
 #2 Local - A simple, narrowly scoped transaction involving many statements but only a single database.
#3 Global - A complex, broadly scoped transaction involving many statements and many databases or potentially other transaction capable resources such as JMS queues or JCA connections.
#4 Custom - iBATIS supports user-provided connections for which you can manage the transactions however you like.

Local transactions are the most common type of transaction, and are really the minimum you should use on any project involving a relational database. Configuration as below for example:

<transactionManager type="JDBC">
<dataSource type="SIMPLE">
<property …/>
<property …/>
<property …/>
</dataSource>
</transactionManager>


The type="JDBC" attribute tells iBATIS to use the standard JDBC Connection API
for managing transactions.


public void runStatementsUsingLocalTransactions() {
SqlMapClient sqlMapClient =
SqlMapClientConfig.getSqlMapClient();
try {
sqlMapClient.startTransaction();
Person p = (Person)sqlMapClient.queryForObject ("getPerson", new Integer(9));
p.setLastName("Smith");
sqlMapClient.update("updatePerson", p);
Department d = (Department)sqlMapClient.queryForObject ("getDept", new Integer(3));
p.setDepartment(d);
sqlMapClient.update("updatePersonDept", p);
sqlMapClient.commitTransaction();
} finally {
sqlMapClient.endTransaction();
}
}


The two update statements above will be run within the same transaction,
and therefore if either one fails, both will fail.