4.4 Using Oracle Managed Files with Exascale

To store Oracle Database files on Exascale, Oracle recommends using the Oracle Managed Files (OMF) feature of Oracle Database.

Using OMF, you can simply define Oracle Database instance parameters (init.ora parameters) that specify the Exascale vault as the target destination. For example:

db_create_file_dest=@MYVAULT
db_create_online_log_dest_1=@MYVAULT
db_recovery_file_dest=@MYVAULT

You can also use OMF when specifying an Oracle Database file name in a SQL command. For example:

SQL> CREATE TABLESPACE mytbs DATAFILE '@MYVAULT' SIZE 10G;

If desired, you can specify OMF file locations that contain a more specific path inside an Exascale vault. For example:

db_create_file_dest=@MYVAULT/mylocation/

Note:

If you specify an OMF file location with a path inside an Exascale vault, you must include a forward slash character (/) at the end of the value.

This requirement also applies when specifying a path inside an Exascale vault in a SQL command or to an Oracle Database client (such as RMAN). For example:

RMAN> set archivelog destination to '@MYVAULT/mylogs/';

Using OMF, Oracle Database files are created and managed automatically using the following file naming convention:

OMF_dest/[cluster_details/]db_unique_name/[pdb_guid_name/]file_type/[date_stamp/]name.OMF.random

In the file naming convention:

• OMF_dest: Depending on the context, OMF_dest specifies the OMF target value specified directly in the SQL command or the value of the relevant OMF instance parameter. For example, if no value is explicitly specified in the SQL command, the value of db_create_file_dest is used for a data file, while the value of db_recovery_file_dest is used for an archived redo log file.

• vault: Specifies the Exascale vault name.

• cluster_details: Identifies the Oracle Grid Infrastructure (GI) cluster that contains the Oracle database. The value uses the format: clustername-clusterGUID.

  To find the clusterGUID for a GI cluster, you can use the following command:

  # crsctl get css clusterguid

  If the Oracle database is not associated with a GI cluster, then this element is not included in the file path.

• db_unique_name: Identifies the Oracle database by using the db_unique_name database parameter value.

• pdb_guid_name: Identifies the pluggable database (PDB). If the Oracle database is not a multitenant container database (CDB), then this element is not included in the file path.

• file_type: Identifies the Oracle Database file type; for example, DATAFILE, ONLINELOG, CONTROLFILE, and so on.

• date_stamp: Specifies the file creation date. This element is only included for specific Oracle Database file types where files are automatically generated over time, such as archive log files and automatically generated backup files.

• name: Specifies the OMF file name. The specific value depends on the Oracle Database file type. For example, a data file name contains the relevant tablespace name, while an archive log file name identifies the log thread and sequence number.

• random: Specifies a random alpha-numeric string to guarantee uniqueness.