You can create a database for the job scheduler and grid endpoint if you do not use the default Apache Derby database. The job scheduler stores job information in a relational database while the grid endpoint uses the database to track the progress of a batch job.
When you install the feature pack, one Derby Java™ Database Connectivity (JDBC) provider is created. The Derby JDBC provider contains two data sources. One is the default Derby data source, JNDI name jdbc/lrsched, that points to the default Derby job scheduler database. The other, JNDI name jdbc/pgc, is the batch execution environment data source. If you decide to use the default data source you do not need to create the job scheduler database. The default Derby database for the job scheduler is created when the job scheduler host (deployment target) is selected through the administrative console. The default Derby database for the endpoint is created when a batch application is first installed on a node. Embedded Derby databases cannot be shared by multiple processes and are unsuitable for environments where the job scheduler must move from one node to another. For example, the job scheduler must move from one node to another in high availability scenarios.
The feature pack supports Derby, DB2®, and Oracle databases. You can use the following steps to configure the job scheduler and grid endpoint database if you decide to use a database other than the Derby database. When you create the database manually, the job scheduler and grid endpoint can use the same database.
Consult the JDBC provider documentation for more information about defining a new JDBC provider.
Define the data source at the cell level. Doing so guarantees that the database is available for each application server that hosts the job scheduler.