Upgrade or Downgrade Oracle Machine Learning for SQL

38.2.1 Pre-Upgrade Steps

Pre-upgrade considerations.

Before upgrading, you must drop any machine learning models and machine learning activities that were created inOracle Data Miner.

38.2.2 Upgrade Oracle Machine Learning for SQL

You can upgrade your database by using the Database Upgrade Assistant (DBUA) or you can perform a manual upgrade using export/import utilities.

All models and machine learning metadata are fully integrated with the Oracle Database upgrade process whether you are upgrading from 19c or from earlier releases.

Upgraded models continue to work as they did in prior releases. Both upgraded models and new models that you create in the upgraded environment can make use of the new machine learning functionality introduced in the new release.

Related Topics

38.2.2.1 Use Database Upgrade Assistant to Upgrade Oracle Machine Learning for SQL

Oracle Database Upgrade Assistant provides a graphical user interface that guides you interactively through the upgrade process.

On Windows platforms, follow these steps to start the Upgrade Assistant:

Go to the Windows Start menu and choose the Oracle home directory.
Choose the Configuration and Migration Tools menu.
Launch the Upgrade Assistant.

On Linux platforms, run the DBUA utility to upgrade Oracle Database.

Related Topics

Oracle Database Upgrade Guide

38.2.2.2 Use Export/Import to Upgrade Machine Learning Models

Use Export and Import functions of the Oracle Database to export the previously created models and import the models in an instance of Oracle Database version.

If required, you can use a less automated approach to upgrading machine learning models. You can export the models created in a previous version of Oracle Database and import them into an instance of the Oracle Database version.

38.2.2.2.1 Export/Import Oracle Machine Learning for SQL Models

Export and import Oracle Machine Learning for SQL models.

To export models from an instance of a previous release of Oracle Database to a dump file, follow the instructions in Export and Import Oracle Machine Learning for SQL Models.

To import the dump file into the Oracle Database database:

%ORACLE_HOME\bin\impdp system\<password> 
       dumpfile=<dumpfile_name> 
       directory=<directory_name> 
       logfile=<logfile_name> .....
SQL>CONNECT / as sysdba;
SQL>EXECUTE dmp_sys.upgrade_models();
SQL>ALTER SYSTEM flush shared_pool;
SQL>ALTER SYSTEM flush buffer_cache;
SQL>EXIT;

ALTER SYSTEM Statement

You can flush the Database Smart Flash Cache by issuing an ALTER SYSTEM FLUSH FLASH_CACHE statement. Flushing the Database Smart Flash Cache can be useful if you need to measure the performance of rewritten queries or a suite of queries from identical starting points.

38.2.3 Post Upgrade Steps

Perform steps to view the upgraded database.

After upgrading the database, check the DBA_MINING_MODELS view in the upgraded database. The newly upgraded machine learning models must be listed in this view.

After you have verified the upgrade and confirmed that there is no need to downgrade, you must set the initialization parameter COMPATIBLE to 21.0.0. In Oracle Database 21c, when the COMPATIBLE initialization parameter is not set in your parameter file, the COMPATIBLE parameter value defaults to 21.0.0.

Note:

The CREATE MINING MODEL privilege must be granted to Oracle Machine Learning for SQL user accounts that are used to create machine learning models.

Related Topics

38.2.4 Downgrade Oracle Machine Learning for SQL

Before downgrading the Oracle database back to the previous version, ensure that no models are present.

Use the DBMS_DATA_MINING.DROP_MODEL routine to drop the models before downgrading. If you do not do this, the database downgrade process terminates.

Issue the following SQL statement in SYS to verify the downgrade:

SQL>SELECT o.name FROM sys.model$ m, sys.obj$ o 
                  WHERE m.obj#=o.obj# AND m.version=2;