Oracle Database Features with Limited Support in Autonomous AI Database on Dedicated Exadata Infrastructure

Autonomous AI Database on Dedicated Exadata Infrastructure supports most of the features present in Oracle Database Enterprise Edition. To ensure the security and the performance of Autonomous AI Database, some features have only limited support.

The following Oracle Database features, options and packs have limited support in Autonomous AI Database.

Feature, Option or Pack	Limitations
Database In-Memory	See Database In-Memory.
Fast Ingest	See Fast Ingest.
Index-organized tables (IOTs)	Index-organized tables are not supported, but attempting to create one does not generate an error. Instead, a heap-organized table with a primary key index is created. Therefore, if you use index-organized tables, you should test the applications that use index-organized tables to confirm that they work using heap-organized tables with a primary key indexes.
Java in Oracle Database (sometimes called OJVM or Database Embedded JVM)	Database sessions that use Java in Oracle Database will experience reduced database-side performance when service maintenance operations are being performed on the database.
LogMiner	See Oracle Log Miner.
Oracle Data Pump	Transportable tablespace mode (the `TRANSPORT_TABLESPACES` parameter) is not supported.
Oracle Database Advanced Queuing	Only the PL/SQL Advanced Queuing APIs are supported. Oracle Call Interface (OCI) notifications for Advanced Queuing are not supported.
Oracle Flashback and Restore Points	See Oracle Flashback and Restore Points.
Oracle Label Security	You cannot specify labeling functions in policies.
Oracle Machine Learning (OML)	Supports: OML4SQL on all database versions. Oracle Data Miner. For OML product documentation, refer to Oracle Machine Learning on Oracle Help Center.
Oracle Real Application Testing	SQL Performance Analyzer and Database Replay are the key components of Oracle Real Application Testing (RAT). Depending on the nature and impact of the system change being tested and the type of test system, you can use either or both components to perform your testing. Database Replay: See Database Replay for more details. SQL Performance Analyzer (SPA): See Testing Guide in Oracle Database 19c or Oracle Database 26ai for more information on SPA. Tip: You can use the SPA features from Enterprise Manager (EM). For more details, see Monitoring and Administration Tasks in Administrator's Guide for Oracle Autonomous AI Databases
Oracle Text	See Oracle Text.
Oracle XML DB	See Oracle XML DB.

SQL Commands

Autonomous AI Database on Dedicated Exadata Infrastructure allows most of the SQL commands available in Oracle Database. To ensure the security and the performance of Autonomous AI Database, some SQL commands are restricted.

This section provides a list of SQL command limitations that are required to protect security and for performance integrity in Autonomous AI Database on Dedicated Exadata Infrastructure. Most of the standard SQL and PL/SQL syntax and constructs used with Oracle Database work in Autonomous AI Database.

Note: If you try to use a restricted SQL command the system generates this error:

ORA-01031: insufficient privileges

This error indicates that you are not allowed to run the SQL command in Autonomous AI Database.

The following SQL commands and PL/SQL packages are not supported in Autonomous AI Databases:

ADMINISTER KEY MANAGEMENT
SYS.DBMS_SHARED_POOL
CREATE_LIBRARY

SQL Commands with Restrictions in Autonomous AI Database

The following DDL commands are available in Autonomous AI Database with restrictions:

SQL Command	Restrictions
`ALTER PLUGGABLE DATABASE` and `ALTER DATABASE`	Only the following clauses are allowed: `DATAFILE AUTOEXTEND ON` `DATAFILE AUTOEXTEND OFF` `DATAFILE RESIZE` `DEFAULT EDITION` `SET TIME_ZONE`
`ALTER PROFILE`	Using `ALTER PROFILE`: There are restrictions for a user defined `PASSWORD_VERIFY_FUNCTION`. See Manage Password Complexity on Autonomous AI Database for more information. The optional `CONTAINER` clause is ignored if specified.
`ALTER SESSION`	Only the following clauses are allowed: `ADVISE COMMIT, ADVISE ROLLBACK, ADVISE NOTHING` `CLOSE DATABASE LINK` `ENABLE COMMIT IN PROCEDURE, DISABLE COMMIT IN PROCEDURE` `ENABLE PARALLEL , DISABLE PARALLEL , FORCE PARALLEL` `ENABLE RESUMABLE, DISABLE RESUMABLE` `SET DEFAULT_COLLATION` `SET EDITION` `SET ISOLATION_LEVEL` `SET ROW ARCHIVAL VISIBILITY` `SET init-param = value` For a list of initialization parameters you can set in Oracle Autonomous AI Database on Dedicated Exadata Infrastructure, see Modifying Database Initialization Parameters. Note that you cannot set some of these parameters at the session level. To find out whether a given parameter can be altered using `ALTER SESSION SET`, query the `ISSES_MODIFIABLE` column of the `V$PARAMETER` dynamic performance view.
`ALTER SYSTEM`	Only `ALTER SYSTEM SET` and `ALTER SYSTEM KILL SESSION` are allowed. `SET` can only be used to set parameters listed in Modifying Database Initialization Parameters.
`ALTER TABLESPACE`	The use of `SMALLFILE` tablespaces is not supported by Autonomous AI Database. The Autonomous AI Database block size is fixed at 8K. A `BIGFILE` tablespace contains only one data file or temp file, and the maximum size of the file for a tablespace with 8K blocks is 32 terabytes (TB).
`ALTER TABLE`	The following clauses are ignored: `allocate_extent` `alter_iot` `deallocate_unused` `ilm` `inmemory_table` `logging` `modify_LOB_storage` `physical_attributes` `shrink` For more information on `ALTER TABLE`, see Oracle Database 19c SQL Language Reference or Oracle AI Database 26ai SQL Language Reference.
`CREATE DATABASE LINK`	In the `USING 'connect string'` clause, you must use the Easy Connect syntax or provide a complete descriptor. You cannot use a network service name because the `tnsnames.ora` file is not available for lookup. Database links can only be used for TCP connections because TCPS connections require a wallet.
`CREATE PROFILE`	Refer to Manage Password Complexity on Autonomous AI Database for the `Password Verify Function` requirements.
`CREATE TABLE`	`XMLType` tables using XML schema-based storage are not allowed. See Oracle XML DB. The following clauses are ignored: `cluster` `ilm_clause` `inmemory_table_clause` `LOB_storage_clause` `logging_clause` `organization external`: Use of this clause has certain limitations, as described in Manually Created External Tables. `organization index`: Using this clause does not create an index-organized table. It creates a regular table with a primary key. You should test and verify the performance of the generated table for your application. `physical_properties` For more information on `CREATE TABLE`, see Oracle Database 19c SQL Language Reference or Oracle AI Database 26ai SQL Language Reference.
`CREATE TABLESPACE`	The use of `SMALLFILE` tablespaces is not supported by Autonomous AI Database. The Autonomous AI Database block size is fixed at 8K. A `BIGFILE` tablespace contains only one data file or temp file, and the maximum size of the file for a tablespace with 8K blocks is 32 terabytes (TB).
`CREATE USER`	The following clauses can not be used with `CREATE USER` statement: `IDENTIFIED EXTERNALLY` `IDENTIFIED BY VALUES`
`DROP TABLESPACE`	This command is supported only when used in the following format: `DROP TABLESPACE INCLUDING CONTENTS AND DATAFILES;` If you execute this command without `INCLUDING CONTENTS AND DATAFILES` in the syntax, you will get the following error: `ORA-01031: insufficient privileges`

Data Types

Autonomous AI Database on Dedicated Exadata Infrastructure allows most of the data types available in Oracle Database. To ensure the security and the performance of Autonomous AI Database, some data types are restricted.

The following Oracle Database data types are not supported or are only partially supported in Autonomous AI Database:

By default, the ROWID and UROWID data types are not enabled in Dedicated Exadata Infrastructure deployments. You cannot specify ROWID or UROWID columns when creating tables, nor can you add ROWID or UROWID columns to existing tables.

However, you can enable ROWID or UROWID by setting the ALLOW_ROWID_COLUMN_TYPE initialization parameter to true. See Database 19c Reference or Database 26ai Reference for more information on enabling it.
Large object (LOB) data types: only SecureFiles LOB storage is supported.
Multimedia data types are not supported. (Oracle Multimedia is desupported in Oracle Database.)

For a list of Oracle data types see Oracle Database 19c SQL Language Reference or Oracle Database 26ai SQL Language Reference.

Database In-Memory

Oracle Database In-Memory (Database In-Memory) is a suite of features that significantly improves performance for real-time analytics and mixed workloads.

The In-Memory Column Store (IM column store) is the key feature of Database In-Memory. The IM column store maintains copies of tables, partitions, and individual columns in a special compressed columnar format optimized for rapid scans. The IM column store resides in the In-Memory Area, which is an optional portion of the system global area (SGA). For more information about this feature, refer to Database 19c In-Memory Guide or Database 26ai In-Memory Guide.

Autonomous AI Database on Dedicated Exadata Infrastructure lets you enable Database In-memory for your Autonomous AI Database by allocating a percentage of its System Global Area (SGA) to the In-Memory column store (IM column store) either while provisioning the database or later. You can also enable or disable Database In-Memory for an existing Autonomous AI Database from its Details page on the Oracle Cloud Infrastructure (OCI) console.

The following describes notes to consider before and after enabling Database In-Memory for your Autonomous AI Database:

Turning Database In-Memory on or off is supported only on Autonomous AI Databases with at least 16 ECPUs or 4 OCPUs enabled.
Scaling down an Autonomous AI Database to less than 16 ECPUs or 4 OCPUs automatically disables Database In-Memory for that database.
You can also enable Database In-Memory on databases cloned from a database instance or backup, irrespective of whether the clone source has Database In-memory enabled. As the settings are not inherited from the clone, you can enable Database In-Memory manually while cloning or later.
You may experience performance degradation in the following cases:
- Any patching activity that results in rebooting the database with Database In-Memory enabled.
- Restarting an Autonomous Container Database with Database In-Memory enabled Autonomous AI Databases.
If you enable Database In-Memory for a primary Autonomous AI Databases in an Autonomous Data Guard setup, the standby database also carries the same settings. However, you can not change these settings on the standby database.
For better management and sharing of the underlying SGA/memory resources, Oracle recommends that all the Autonomous AI Databases configured for Database In-Memory be in the same Autonomous Container Database.

Refer to Enable or Disable Database In-Memory or Create an Autonomous AI Database on Dedicated Exadata Infrastructure for detailed instructions.

Manually Created External Tables

Instead of using the DBMS_CLOUD.CREATE_EXTERNAL_TABLE procedure to create an external table, you can use the SQL CREATE TABLE...ORGANIZATION EXTERNAL statement to manually create one. When you do so, certain limitations apply.

The ORACLE_LOADER and ORACLE_BIGDATA access drivers are supported. The ORACLE_DATAPUMP, ORACLE_HDFS and ORACLE_HIVE access drivers are not supported.
If you are specifying files stored in Cloud Object Storage as the data source location, you must use one of the URI formats described in Cloud Object Storage URI Formats.

Oracle LogMiner

LogMiner, which is part of Oracle Database, enables you to query online and archived redo log files through a SQL interface. Redo log files contain information about the history of activity on a database.

All changes made to user data or the database dictionary are recorded in the Oracle redo log files to perform database recovery operations. Because LogMiner provides a well-defined, easy-to-use, and comprehensive relational interface to redo log files, it can be used as a powerful data auditing tool and sophisticated data analysis tool. For key capabilities of LogMiner, refer to LogMiner Benefits in Oracle Database 19c Utilities or Oracle AI Database 26ai Utilities .

Autonomous AI Database on Dedicated Exadata Infrastructure supports LogMiner, with the following limitations:

The DBMS_LOGMNR.ADD_LOGFILE and DBS_LOGMNR.REMOVE_LOGFILE procedures can not be used from an Autonomous AI Database.
The dbms_logmnr.start_logmnr procedure can only be invoked with SCN or time range parameters. Querying the V$LOGMNR_CONTENTS view subsequently returns the data from the specified SCN or time range for the connected Autonomous AI Database.
Only one LogMiner session is allowed per Autonomous AI Database.
By default, Autonomous AI Database archived log files are kept for up to:
- 3 days for databases without GoldenGate capture and Autonomous Data Guard setup.
- 7 days for databases with Autonomous Data Guard that do not have GoldenGate capture enabled.
- 15 days for databases with GoldenGate capture enabled.
Therefore, LogMiner can only access up to 3, 7, or 15 days of archived log files as per the above list.

The log files retention period depends on the database transaction rate and is not always guaranteed up to the number of days specified above. There can be instances when you attempt to mine unavailable log files. In such cases, LogMiner reports ORA-1285: "error reading file".

Note: You must have the GGADMIN role to run LogMiner. See Unlock GGADMIN user for more details.

Oracle Text

The following describes support for Oracle Text features in Autonomous AI Database on Dedicated Exadata Infrastructure. To ensure the security and the performance of Autonomous AI Database, some Oracle Text features are restricted.

Oracle Text Feature	Supported in Autonomous AI Database	More Information
All logging, and APIs which perform logging such as `ctx_report.query_log_summary`	Not Supported	See QUERY_LOG_SUMMARY in Oracle Database 19c or Oracle Database 26ai
File and URL datastore	Not Supported	See Datastore Type in Oracle Database 19c or Oracle Database 26ai
`CREATE INDEX` with `BIG_IO` option	Not supported by default. (see Footnote 1)	See Improved Response Time Using the BIG_IO Option of CONTEXT Index in Oracle Database 19c or Oracle Database 26ai
`OPTIMIZE_INDEX` in rebuild mode	Not supported by default.(see Footnote 1).	OPTIMIZE_INDEX in Oracle Database 19c or Oracle Database 26ai

Footnote 1 This is supported if you grant the privilege to create a trigger to the user (GRANT CREATE TRIGGER). You must also disable parallel DML at the session level (ALTER SESSION DISABLE PARALLEL DML).

For details on Oracle Text, see Oracle Text 19c Application Developer’s Guide or Oracle Text 26ai Application Developer’s Guide.

Oracle XML DB

The following describes support for Oracle XML DB features in Autonomous AI Database on Dedicated Exadata Infrastructure. To ensure the security and the performance of Autonomous AI Database, some Oracle XML DB are restricted.

Note: If you migrate tables containing XMLType columns to Autonomous AI Database on Dedicated Exadata Infrastructure using Oracle Data Pump, you need to convert to Non-Schema Binary XML prior to using Oracle Data Pump Export (expdp).

The following features are supported:

Full support for XMLQuery, XMLTable, and other SQL/XML standard functions
Indexing schema including functional indexes using SQL/XML expressions, Structured XMLIndex and XQuery Full Text Index

The following features are not supported or are only partially supported.

Area	XML DB Feature	Supported in Autonomous AI Database	More Information
Repository	XML DB Protocol	No	See Repository Access Using Protocols in Oracle Database 19c or Oracle Database 26ai
Repository	XML DB Resources	No	See Oracle XML DB Repository Resources in Oracle Database 19c or Oracle Database 26ai
Repository	XML DB ACLs	No	See Repository Access Control in Oracle Database 19c or Oracle Database 26ai
Storage	XML Schema Registration	No	See XML Schema Registration with Oracle XML DB 19c or Oracle XML DB 26ai
Storage	CLOB	No	Deprecated
Storage	Object Relational	No	See XML Schema and Object-Relational XMLType in Oracle Database 19c or Oracle Database 26c
Storage	Binary XML	Yes (Non schema-based only)	See XMLType Storage Models in Oracle Database 19c or Oracle Database 26ai
Index	Structured XML Index	Yes	See XMLIndex Structured Component in Oracle Database 19c or Oracle Database 26ai
Index	XQuery Full Text Index	Yes	See Indexing XML Data for Full-Text Queries in Oracle Database 19c or Oracle Database 26ai
Index	Unstructured XMLIndex	No	See XMLIndex Unstructured Component in Oracle database 19c or Oracle Database 26ai
Packages	XML DOM package	Yes	See PL/SQL DOM API for XMLType (DBMS_XMLDOM) in Oracle Database 19c or Oracle Database 26ai
Packages	XML Parser Package	Yes	See PL/SQL Parser API for XMLType (DBMS_XMLPARSER) in Oracle Database 19c or Oracle Database 26ai
Packages	XSL Processor (DBMS_XSLPROCESSOR)	Yes	See PL/SQL XSLT Processor for XMLType (DBMS_XSLPROCESSOR) in Oracle Database 19c or Oracle Database 26ai

For details on Oracle XML DB, see Oracle XML DB 19c Developer’s Guide or Oracle XML DB 26ai Developer’s Guide.

Oracle Flashback and Restore Points

The following describes support for Oracle Flashback and Restore Points in Autonomous AI Database on Dedicated Exadata Infrastructure. To ensure the security and the performance of Autonomous AI Database, some Oracle Flashback and Restore Point features are restricted.

To restore and recover your database to a point in time, see Restore and Recover.

Flashback and Restore Points Feature	Supported in Autonomous AI Database on Dedicated Exadata Infrastructure
`DBMS_FLASHBACK`	Yes, except for the `DBMS_FLASHBACK.TRANSACTION_BACKOUT` procedure.
Flashback Time Travel	Yes, except for specifying a default Flashback Data Archive using the `SET DEFAULT` clause with the `CREATE FLASHBACK ARCHIVE` and `ALTER FLASHBACK ARCHIVE` statements.
Flashback Drop	Yes
Flashback Query	Yes
Flashback Table	Yes
Flashback Transaction	Yes
Flashback Transaction Query	Yes
Flashback Version Query	Yes
Restore Point	No
Guaranteed Restore Point	No

For information on using Flashback features, see About Oracle Database 19c Flashback Technology or About Oracle Database 26ai Flashback Technology.

Database Replay

You can use Database Replay to capture a workload on the production system and replay it on a test system with the original workload’s exact timing, concurrency, and transaction characteristics. This enables you to test the effects of a system change without affecting the production system.

Capturing the production workload eliminates the need to develop simulation workloads or scripts, resulting in significant cost reduction and time savings. Realistic testing of complex applications that took months using load simulation tools can be completed within days with the help of Database Replay. This enables you to rapidly test changes and adopt new technologies with higher confidence and at lower risk. For more information about Database Replay, see Introduction to Database Replay in Oracle 19c Testing Guide or Oracle 26ai Testing Guide.

Using the Database Replay component of Real Application Testing (RAT) in the Oracle Autonomous AI Database on Dedicated Exadata Infrastructure is the same as its implementation in the Oracle Database. See Database Replay in Oracle Database 19c or Oracle Database 26ai for the process involved in using Database Replay with your database.

There are certain limitations and recommendations to use Database Replay with Autonomous AI Database on Dedicated Exadata Infrastructure:

Oracle recommends that you use a Network File System (NFS) or File Storage System (FSS) that is attached to your Autonomous AI Database to store the capture and replay files. Refer to Attach Network File System and Attach External File System for guidance.
For Database Replay, the replay client (wrc) can be the same version as the replay database, or a version lower to the replay database. However, the minimum version supported for the replay client is 19.24.
Autonomous AI Database does not support dbms_workload_capture.import_awr and dbms_workload_replay.import_awr. That is, you can not import Automatic Workload Repository (AWR) data with Database Replay. Hence, it is recommended that you generate the AWR comparison reports on a non-autonomous database, preferably Oracle Database 26c instance for better compatibility.

Database PL/SQL Packages

The following describes notes to consider before, during, and after using PL/SQL packages in Autonomous AI Database on Dedicated Exadata Infrastructure.

Lists Oracle Database PL/SQL packages with restrictions and notes in Autonomous AI Database on Dedicated Exadata Infrastructure.

Removed PL/SQL Packages

DBMS_DEBUG_JDWP
DBMS_DEBUG_JDWP_CUSTOM
UTL_TCP

Note: While the UTL_TCP package is not supported on Autonomous AI Database on Dedicated Exadata Infrastructure, the UTL_TCP.CRLF constant can be used in specific instances with other packages. See SMTP Send Email Sample Code for an example of using UTL_TCP.CRLF.

PL/SQL Packages Notes

UTL_HTTP Notes
- Connections through IP addresses are allowed.
- Egress ports 80 and 443 are enabled by default.
- HTTP, HTTPS, and HTTP_PROXY connections are allowed.
- The APEX_WEB_SERVICE PL/SQL package is layered on UTL_HTTP.
- Before invoking the UTL_HTTP subprograms, you must first set the wallet location using the SET_WALLET subprogram. As this is a session level setting, you must set it before each session. For more information, see UTL_HTTP in Oracle Database 19c PL/SQL Packages and Types Reference or Oracle Database 26ai PL/SQL Packages and Types Reference.
```
BEGIN
  UTL_HTTP.SET_WALLET(path => 'file:/var/opt/oracle/dbaas_acfs/grid/tcps_wallets');
END;
/
```
UTL_SMTP Notes
- Connections through IP addresses are allowed.
- Egress ports 25, 465, 587, and 2525 are enabled by default.
- Both simple and secure (SSL) SMTP endpoints are accessible.
- Oracle Cloud Infrastructure (OCI) Email Delivery Service and third-party SMTP servers are both supported as email providers. See Oracle Cloud Infrastructure - Networking and Connectivity for details on establishing connectivity between the third-party server and your Autonomous AI Database tenant.
- However, unlike Oracle Enterprise Edition on-premises database, you cannot set up localhost as the SMTP server. The sendmail() service is also disabled on the platform.
- The APEX_MAIL PL/SQL package is layered on UTL_SMTP.
DBMS_AQADM Notes
- DBMS_AQADM.MOVE_QUEUE_TABLE can only be used in ONLINE mode to move Advanced Queuing (AQ) tables from one tablespace to another or within the same tablespace. See My Oracle Support (MOS) Note: Doc ID 2867445.1 for guidance on moving AQ tables across tablespaces using DBMS_AQADM.MOVE_QUEUE_TABLE on 19c and above.
- DBMS_AQADM.MOVE_QUEUE_TABLE can not be used to move AQ tables even in ONLINE mode when table or schema names are case-sensitive and include special identifiers.