The following sections describe Oracle setup tasks.
If you plan to install and use Connect CDC in a configuration where Connect CDC Director will run on a different host than the one on which Oracle is installed, Connect CDC requires the use of the sys user ID. The sys user ID must have privileges to create a database and is used only when you install the SQL metabase. Oracle requires:
-
The sys user ID to be logged in with SYSDBA privileges.
-
A password file when a user logs in remotely with SYSDBA privileges. If you do not have a password file, it must be created before you start modeling in Connect CDC Director.
-
Shut down and restart Oracle when you create a password file.
The database initialization parameter OPEN_CURSORS must be defined at a minimum of 50 if any Replication or Synchronization request contains a bidirectional mapping to an Oracle server. If the number is less than 50, you may receive the Oracle error message indicating you have exceeded the maximum number of cursors if you have many bidirectional mappings to an Oracle server.
Before you use the Change Selector
You must do the following setup activities before you can use the Oracle Change Selector:
-
Archive mode
Running in archive mode is a requirement for replication. If archive mode is not enabled during startup, the Change Selector issues an error and shuts down.
Additionally, the automatic archival option must be enabled by setting log_archive_start = TRUE in the Oracle parameter file and restarting the database. Connect CDC cannot do this automatically because Oracle does not provide a way to do this programmatically.
-
Storage requirements
The log-based changed data capture requires supplemental logging. Extra storage may be required for supplemental logging depending on a number of factors such as, the number of tables being replicated, the number of columns replicated within those tables, the size of the rows within the table and the overall volume of updates against replicated tables.
The best practice is to monitor the actions on the tables and to determine if the database continually requests more archive space. If so, then storage must be increased.
|
Statement |
What happens to the redo logs |
|---|---|
|
Inserts |
No additional information is written to the redo logs. |
|
Deletes |
Additional information, which includes the old column for all replicated columns, is written to the redo logs. Without supplemental logging, normally only the old values for the key columns are written to the redo log. As a result of this additional information in the redo logs, more archive logs may be generated. These archive logs must be kept until they are no longer needed for replication and have been saved for backup. |
|
Updates |
Additional information, which includes the old column for all replicated columns, is written to the redo logs. Without supplemental logging, normally only the old values for the key columns are written to the redo log. As a result of this additional information in the redo logs, more archive logs may be generated. These archive logs must be kept until they are no longer needed for replication and have been saved for backup. |
-
Privileges
-
If you plan to retain archive logs (see step 5), the user who starts the kernel must have the privileges to create files in the primary Oracle archive destination directory.
This privilege is only required when you specify that you want to retain archive logs, as specified for the Oracle parameters during the install. For each processed archive log, Connect CDC creates an empty file using the same archive name with the “.repdone” suffix appended to the name as a means to determine which logs Connect CDC has processed. These logs can thus be backed up and deleted.
-
The user who is to use “Delete unneeded archive logs” (see step 5) needs delete privileges on Oracle’s archive log files.
When delete archive logs is specified Connect CDC does not create the empty files. However the user must have the privileges to delete the archive log files.
-
If you plan to use Oracle’s fast loader, you must specify Oracle’s bin directory in your path. For information on the fast loader, see the Connect CDC Advanced User Guide .