IBM E02HMLL-I Implementation Guide - Page 159

v It is easier to add new records (such as a new routing value paired with a new destination application) when using a database connection because you only have to add a new row to the database; no components have to be modified, and the new pairing is effectively immediate. Adding new associations in Java code, however, is less convenient because you must modify the code of the component, recompile it, and possibly stop and restart it (if the component is a collaboration). You should pause the components of the interface beforehand as well, to ensure that there no transactions are in progress while the component is being compiled, or they may fail. v Performing a lookup in Java code should be faster than making a connection to a database and performing a query, though this depends on how many associations there are that the Java control structure must iterate through. v It is less convenient to migrate an interface that uses a database connection to another environment than it is to migrate an interface that exclusively relies on Java code, because you must remember to migrate not only the map or collaboration that makes the connection, but must also remember to recreate the tables involved in the new environment. Persisting information Some customers want to make information about the operation of the business integration system persistent by storing that information in a database so that it can be referenced for problem resolution or historical analysis. To satisfy this requirement, do the following: 1. Create a database table with as many columns as necessary to store the desired data. 2. Create a database connection pool and database connection in System Manager as described in "Creating database connection pools and database connections" on page 148. 3. In the collaboration template design the logic to do obtain a connection to the database and execute a SQL query that inserts the desired data into the appropriate column. Typically this requirement involves persisting information contained in the business object being processed by the collaboration (such as the primary key of the entity being processed), or information about the system itself (such as the successful processing of a business object request). You can persist business object data like primary key values by using the Business Object Probe feature; for more information, see the Collaboration Development Guide. When not to use database connection pools You should not use database connection pools to connect directly to the database of an application. To interact with an application database you should only use an adapter. Database connections should only be made to databases that do not support an application because adapters use the business logic provided by programming interfaces of the application. If you connect directly to an application database to perform a SQL update statement, for example, then you circumvent any related logic that the API would have performed in response to an update operation. This violates the integrity of the application and the business process. Chapter 8. Configuring database connection pools 147