Administration Guide › System Tasks › How to Add Users with a Feeder File › Bulk Loader Considerations

Bulk Loader Considerations

You can use the Bulk Loader tab to upload feeder files that are used to manipulate large numbers of managed objects simultaneously. Note the following considerations when using the Bulk Loader:

• Consider scheduling large bulk loads during non-peak times, such as overnight. Large bulk loads can affect performance. In some cases, a bulk load that includes many sub tasks can prevent user submitted tasks from completing until the bulk load finishes.
• If the server goes down during a long-running task, such as uploading many objects, restart the task under View Submitted Tasks. When the task restarts, it begins from the last successfully executed record.
• If you are using LDAP as a user store with Solaris, the Bulk Loader can hang during an import. To fix this issue, see the Specify LDAP Connection Settings topic in the Configuration Guide and apply the settings that are outlined there.
• If you use the Bulk Loader to import a large number of users, you may see out-of-memory exceptions. To address this issue, tune the following heap size memory parameters:
  • -Xmx
  • -XX:maxPermSize

    Note: For more information about tuning memory parameters, see your application server documentation.

• Using the bulk loader to manipulate many managed objects, such as creating many users, can affect performance. To improve performance, consider the following recommendations:
  • Divide one large CSV file into multiple small files when using the User Console to do bulk loads. For example, doing ten bulk loads of 10,000 users is quicker than doing one bulk load of 100,000 users.

    Note: A CSV file with more than 50,000 entries may cause issues in the system.

  • Limit the number of tasks that the user who is performing the bulk load has. For example, performance improves when the administrator who initiates the bulk load has only a few tasks. When the administrator has many tasks, CA IdentityMinder has to do more extensive permissions checks, which can affect performance.
  • Limit the number of Policy Xpress policies, which are associated with the bulk changes, that involve provisioning. Also, consider creating simple Policy Xpress policies, which do not affect performance as much as complex policies during a bulk load operation.
  • Verify that you have sufficient system resources.