Preparing the Database for Oracle GoldenGate

1 Preparing the Database for Oracle GoldenGate

Learn how to prepare your database for Oracle GoldenGate, including how to configure connections and logging, how to enable Oracle GoldenGate in your database, how to set the flashback query, and how to manage server resources.

Topics:

Enabling Oracle GoldenGate in the Database
The database services required to support Oracle GoldenGate capture and apply must be enabled explicitly for all Oracle database versions. This is required for Extract and all Replicat modes.
Configuring Connections for Extract and Replicat Processes
Extract and Replicat require a dedicated server connection in the tnsnames.ora file.
Setting Flashback Query
To process certain update records, Extract fetches additional row data from the source database.
Managing Server Resources
Extract interacts with an underlying logmining server in the source database and Replicat interacts with an inbound server in the target database. This section provides guidelines for managing the shared memory consumed by the these servers.
Ensuring Row Uniqueness in Source and Target Tables
Oracle GoldenGate requires a unique row identifier on the source and target tables to locate the correct target rows for replicated updates and deletes.
Configuring Logging Properties
Oracle GoldenGate relies on the redo logs to capture the data that it needs to replicate source transactions.
Using the Extract Automated Capture Mode
The automated capture (auto capture) mode allows automatically capturing the tables that have been enabled for Oracle GoldenGate auto capture.
Additional Oracle GoldenGate Configuration for Your Database
This chapter contains additional configuration considerations that may apply to your database environment.

Enabling Oracle GoldenGate in the Database

The database services required to support Oracle GoldenGate capture and apply must be enabled explicitly for all Oracle database versions. This is required for Extract and all Replicat modes.

To enable Oracle GoldenGate, set the following database initialization parameter. All instances in Oracle RAC must have the same setting.

ENABLE_GOLDENGATE_REPLICATION=true

This parameter alters the DBA_FEATURE_USAGE_STATISTICS view. For more information about this parameter, see Initialization Parameters.

Parent topic: Preparing the Database for Oracle GoldenGate

Configuring Connections for Extract and Replicat Processes

Extract and Replicat require a dedicated server connection in the tnsnames.ora file.

Before you begin, make sure that there is a dedicated user on the Oracle database side, with the required privileges. See Granting User Privileges for Oracle Database 21c and Lower.

On the Oracle GoldenGate side, you direct the Extract and Replicat processes to use these connections by specifying the values for USERID or USERIDALIAS parameter in the Extract and Replicat parameter files.

The following are the security options for specifying the connection string in the Extract or Replicat parameter file.

Credential store method:

USERIDALIAS ggeast

In the case of USERIDALIAS, the alias ggeast is stored in the Oracle GoldenGate credential store with the actual connection string. The following example uses the INFO CREDENTIALSTORE command to display the details of the credentials configured in Oracle GoldenGate:

INFO CREDENTIALSTORE DOMAIN OracleGoldenGate

Output:

Domain: OracleGoldenGate
  Alias: ggeast
  Userid: ggadmin@dc1.example.com:1521/DBEAST.example.com

Setting up a Bequeath connection

Oracle GoldenGate can connect to a database instance without using the network listener if a Bequeath connect descriptor is added in the tnsnames.ora.

The following example shows the configuration for connecting to a database using Bequeath connect descriptor:

dbbeq =  (DESCRIPTION=   
      (ADDRESS=(PROTOCOL=beq)     
          (ENVS='ORACLE_SID=sales,ORACLE_HOME=/app/db_home/oracle,LD_LIBRARY_PATH=/app/db_home/oracle/lib')     
          (PROGRAM=/app/db_home/oracle/bin/oracle)     
      (ARGV0=oraclesales)     
      (ARGS='(DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))'))   
        (CONNECT_DATA=(SID=sales)))

In this example:

/app/db_home is the target Oracle database installation directory

sales is the database service name

The ORACLE_SID, ORACLE_HOME, and LD_LIBRARY_PATH in the ENVS parameter refers to the target.

Note:

Make sure that there is no white space between these environment variable settings.

Granting User Privileges for Oracle Database 21c and Lower

Parent topic: Preparing the Database for Oracle GoldenGate

Granting User Privileges for Oracle Database 21c and Lower

The user privileges that are required for connecting to Oracle database from Oracle GoldenGate depend on the type of user.

Privileges should be granted depending on the actions that the user needs to perform as the GoldenGate Administrator User on the source and target databases. For example, to grant DML operation privileges to insert, update, and delete transactions to a user, use the GRANT ANY INSERT/UPDATE/DELETE privileges and to further allow users to work with tables and indexes as part of DML operations, use the GRANT CREATE/DROP/ALTER ANY TABLE/INDEX privileges.

If the GoldenGate Administrator user has the DBA role, additional object privileges are not needed. However, there might be security constraints granting the DBA role to the GoldenGate Administration user. The DBA role is not necessarily required for Oracle GoldenGate.

If there are many objects being replicated, you might consider using the ANY privilege for DML and DDL operations. This simplifies the provision of privileges to the GoldenGate Administrator users, as you only need to grant a few privileges depending on the database operations.

The following table describes some of the essential privileges for GoldenGate Administrator user for Oracle database. For explanation purposes, the table uses c##ggadmin as an example of a common user for a multitenant container database and ggadmin as the pluggable database (PDB) user. PDBEAST and PDBWEST are used as examples of PDB names.

The following table describes the essential privileges for GoldenGate Administrator user for using Oracle GoldenGate with on source and target Oracle databases:

Privilege	Extract	Replicat All Modes	Purpose
`RESOURCE`	Yes	Yes	Required to create objects In Oracle Database 12cR1 and later, instead of `RESOURCE`, grant the following privilege: `ALTER USER user QUOTA {size \| UNLIMITED} ON tablespace;`
`CONNECT`	Yes	Yes	Common user `SYSTEM` connects to the root container. This privilege is essential when the DBA role is not assigned to the user. See an example of Example: Grant privileges using the DBMS_GOLDENGATE_AUTH.GRANT_ADMIN_PRIVILEGE package.
`CREATE SESSION`	Yes	Yes	Required to connect to the database.
`CREATE VIEW`	Yes	Yes	Required to add the heartbeat table view. If you want to be specific to each object, you can also provide the privileges for each object individually. You may consider creating a specific database role to maintain such privileges.
`ALTER SYSTEM`	Yes	Yes	Perform administrative changes, such as enabling logging.
`ALTER USER`	Yes	Yes	Required for multitenant architecture and `GGADMIN` should be a valid Oracle GoldenGate administrator schema.
`EXEC DBMS_GOLDENGATE_AUTH.GRANT_ADMIN_PRIVILEGE ('REPUSER', CONTAINER=>'PDBEAST');`	Yes	Yes	Required for Oracle Autonomous Database Extract and Replicat. Extracts in the root container (`CDB$ROOT)`) might require a value of `ALL` or a specific PDB (example: `pdbeast`). Grant privileges for Extract and Replicat users. See Example: Grant privileges using the DBMS_GOLDENGATE_AUTH.GRANT_ADMIN_PRIVILEGE package. Grant privilges to capture from Virtual Private Database Grants privilges to capture redacted data
Grant `DV_GOLDENGATE_ADMIN` and `DV_GOLDENGATE_REDO_ACCESS` privileges connected as `SYS` user to the Extract and the Replicat user.	Yes	Yes	Capture from Data Vault. See Privileges for Capturing from Oracle Data Vault.
Grant Replicat privileges in `DBMS_MACADM.ADD_AUTH_TO_REALM` if applying to a realm.	NA	Yes	Capture from Data Vault. See Privileges for Capturing from Oracle Data Vault.
`INSERT`, `UPDATE`, `DELETE` on target tables	NA	Yes	Apply replicated DML to target objects. See Details of Support for Objects and Operations in Oracle DML
`GRANT INSERT ANY TO...` `GRANT UPDATE ANY TO...` `GRANT DELETE ANY TO...`	NA	Yes	Grant these privileges to the Replicat user, instead of granting `INSERT`, `UPDATE`, `DELETE` to every table, if replicating every table.
If DDL replication is performed, grant the following as Database Vault owner: `EXECUTE DBMS_MACADM.AUTHORIZE_DDL(‘GGADMIN USER', ‘SCHEMA FOR DDL’);`	No	No	Capture from Data Vault. See Privileges for Capturing from Oracle Data Vault.
DDL privileges on target objects (if using DDL support)	NA	Yes	Issue replicated DDL on target objects. See Details of Support for Objects and Operations in Oracle DDL.
`GRANT [CREATE\|ALTER\|DROP] ANY [TABLE\|INDEX\|VIEW\|PROCEDURE] to GGADMIN;`	Yes	Yes	Grants privileges for DDL Replication for tables.
`CREATE ANY TABLE`	Yes	Yes	Grants privileges for creating table in any schema. To allow creating tables only in a specific schema, use the `CREATE TABLE` privilege.
`CREATE ANY VIEW`	Yes	Yes	Grants privilges to create view in any database schema. To allow creating views in a specific schema, use the `CREATE VIEW` privilege.
`SELECT ANY DICTIONARY`	Yes	Yes	Allow all privileges to work properly on dictionary tables.

Example: Permissions granted for the Oracle database common user

Privilges granted for the Oracle database common user, which is c##ggadmin in the following example:

CREATE USER c##ggadmin IDENTIFIED BY passw0rd CONTAINER=all DEFAULT
TABLESPACE GG_DATA TEMPORARY TABLESPACE temp;
GRANT RESOURCE to c##ggadmin;
GRANT CREATE SESSION to c##ggadmin;
GRANT CREATE VIEW to c##ggadmin;
GRANT CREATE TABLE to c##ggadmin;
GRANT CONNECT to c##ggadmin CONTAINER=all; 
GRANT DV_GOLDENGATE_ADMIN; –-- for data vault user
GRANT DV_GOLDENGATE_REDO_ACCESS; –-- for data vault user
GRANT ALTER SYSTEM to c##ggadmin;
GRANT ALTER USER to c##ggadmin;
ALTER USER c##ggadmin SET CONTAINER_DATA=all CONTAINER=current;
ALTER USER c##ggadmin QUOTA unlimited ON GG_DATA;
GRANT SELECT ANY DICTIONARY to c##ggadmin;
GRANT SELECT ANY TRANSACTION to c##ggadmin;
EXEC DBMS_GOLDENGATE_AUTH.GRANT_ADMIN_PRIVILEGE('c##ggadmin');

In this example, DBA privilege is not provided but the user will be able to access the DBA_SYS_PRIVS package, if required.

Privileges granted for PDB user ggadmin are provided in the following example:

Example: Grant privileges using the DBMS_GOLDENGATE_AUTH.GRANT_ADMIN_PRIVILEGE package

This procedure grants the privileges needed by a user to be an Oracle GoldenGate administrator The following example grants explicit privileges for Extract on Oracle multitenant database:

BEGIN  
DBMS_GOLDENGATE_AUTH.GRANT_ADMIN_PRIVILEGE 
(GRANTEE => 'c##ggadmin',  PRIVILEGE_TYPE => 'CAPTURE',  GRANT_SELECT_PRIVILEGES => TRUE,  DO_GRANTS => TRUE,  CONTAINER => 'ALL'  ); 
END;

See DBMS_GOLDENGATE_AUTH in Oracle Database PL/SQL Packages and Types Reference for more information.

Privileges for Capturing from Oracle Data Vault

Parent topic: Configuring Connections for Extract and Replicat Processes

Privileges for Capturing from Oracle Data Vault

Grant the following privileges connected as SYS user in Oracle database. These privileges are set for Extract and Replicat user credentials:

EXEC DBMS_GOLDENGATE_AUTH.GRANT_ADMIN_PRIVILEGE ('userID','*',
GRANT_OPTIONAL_PRIVILEGES=>'*');
GRANT DV_GOLDENGATE_ADMIN, DV_GOLDENGATE_REDO_ACCESS to userID;

Grant Replicat the privileges in DBMS_MACADM.ADD_AUTH_TO_REALM if applying to a realm.

Connect as Database Vault owner and execute the following scripts:


BEGIN
DVSYS.DBMS_MACADM.ADD_AUTH_TO_REALM(
REALM_NAME => 'Oracle Default Component Protection Realm',GRANTEE => 'userID',AUTH_OPTIONS => 1) ;
END ;
/
EXECUTE_DBMS_MACADM.AUTHORIZE_DDL('SYS', 'SYSTEM');

For DDL replication, grant the following as the Database Vault owner:
```
EXECUTE DBMS_MACADM.AUTHORIZE_DDL
(‘userID', ‘SCHEMA FOR DDL’);
```

Parent topic: Granting User Privileges for Oracle Database 21c and Lower

Setting Flashback Query

To process certain update records, Extract fetches additional row data from the source database.

Oracle GoldenGate fetches data for the following:

User-defined types
Nested tables
XMLType objects

By default, Oracle GoldenGate uses Flashback Query to fetch the values from the undo (rollback) tablespaces. That way, Oracle GoldenGate can reconstruct a read-consistent row image as of a specific time or SCN to match the redo record.

For best fetch results, configure the source database as follows:

Set a sufficient amount of redo retention by setting the Oracle initialization parameters UNDO_MANAGEMENT and UNDO_RETENTION as follows (in seconds).
```
UNDO_MANAGEMENT=AUTO

UNDO_RETENTION=86400 

UNDO_RETENTION can be adjusted upward in high-volume environments.
```
Calculate the space that is required in the undo tablespace by using the following formula.
```
undo_space = UNDO_RETENTION * UPS + overhead
```
Where:
- undo_space is the number of undo blocks.
- UNDO_RETENTION is the value of the UNDO_RETENTION parameter (in seconds).
- UPS is the number of undo blocks for each second.
- overhead is the minimal overhead for metadata (transaction tables, etc.).
Use the system view V$UNDOSTAT to estimate UPS and overhead.
For tables that contain LOBs, do one of the following:
- Set the LOB storage clause to RETENTION. This is the default for tables that are created when UNDO_MANAGEMENT is set to AUTO.
- If using PCTVERSION instead of RETENTION, set PCTVERSION to an initial value of 25. You can adjust it based on the fetch statistics that are reported with the STATS EXTRACT command. If the value of the STAT_OPER_ROWFETCH CURRENTBYROWID or STAT_OPER_ROWFETCH_CURRENTBYKEY field in these statistics is high, increase PCTVERSION in increments of 10 until the statistics show low values.

Grant either of the following privileges to the Oracle GoldenGate Extract user:

GRANT FLASHBACK ANY TABLE TO db_user

GRANT FLASHBACK ON schema.table TO db_user

Oracle GoldenGate provides the following parameters to manage fetching.

Parameter or Command	Description
`STATS EXTRACT` command with `REPORTFETCH` option	Shows Extract fetch statistics on demand.
`STATOPTIONS` parameter with `REPORTFETCH` option	Sets the `STATS EXTRACT` command so that it always shows fetch statistics.
`MAXFETCHSTATEMENTS` parameter	Controls the number of open cursors for prepared queries that Extract maintains in the source database, and also for `SQLEXEC` operations.
`MAXFETCHSTATEMENTS` parameter	Controls the default fetch behavior of Extract: whether Extract performs a flashback query or fetches the current image from the table.
`FETCHOPTIONS` parameter with the `USELATESTVERSION` or `NOUSELATESTVERSION` option	Handles the failure of an Extract flashback query, such as if the undo retention expired or the structure of a table changed. Extract can fetch the current image from the table or ignore the failure.
`REPFETCHEDCOLOPTIONS` parameter	Controls the response by Replicat when it processes trail records that include fetched data or column-missing conditions.

Parent topic: Preparing the Database for Oracle GoldenGate

Managing Server Resources

Extract interacts with an underlying logmining server in the source database and Replicat interacts with an inbound server in the target database. This section provides guidelines for managing the shared memory consumed by the these servers.

The shared memory that is used by the servers comes from the Streams pool portion of the System Global Area (SGA) in the database. Therefore, you must set the database initialization parameter STREAMS_POOL_SIZE high enough to keep enough memory available for the number of Extract and Replicat processes that you expect to run in integrated mode. Note that Streams pool is also used by other components of the database (like Oracle Streams, Advanced Queuing, and Datapump export/import), so make certain to take them into account while sizing the Streams pool for Oracle GoldenGate.

By default, one Extract requests the logmining server to run with MAX_SGA_SIZE of 1GB. Thus, if you are running three Extracts in the same database instance, you need at least 3 GB of memory allocated to the Streams pool. As a best practice, keep 25 percent of the Streams pool available. For example, if there are 3 Extracts, set STREAMS_POOL_SIZE for the database to the following value:

3 GB * 1.25 = 3.75 GB

Parent topic: Preparing the Database for Oracle GoldenGate

Ensuring Row Uniqueness in Source and Target Tables

Oracle GoldenGate requires a unique row identifier on the source and target tables to locate the correct target rows for replicated updates and deletes.

Unless a KEYCOLS clause is used in the TABLE or MAP statement, Oracle GoldenGate selects a row identifier to use in the following order of priority, depending on the number and type of constraints that were logged (see Configuring Logging Properties).

Primary key if it does not contain any extended (32K) VARCHAR2/NVARCHAR2 columns. Primary key without invisible columns.
Unique key: Unique key without invisible columns.

In the case of a non-integrated Replicat, the selection of the unique key is as follows:
- First unique key alphanumerically with no virtual columns, no UDTs, no function-based columns, no nullable columns, and no extended (32K) VARCHAR2/NVARCHAR2 columns. To support a key that contains columns that are part of an invisible index, you must use the ALLOWINVISIBLEINDEXKEYS parameter in the Oracle GoldenGate GLOBALS file.
- First unique key alphanumerically with no virtual columns, no UDTs, no extended (32K) VARCHAR2/NVARCHAR2 columns, or no function-based columns, but can include nullable columns. To support a key that contains columns that are part of an invisible index, you must use the ALLOWINVISIBLEINDEXKEYS parameter in the Oracle GoldenGate GLOBALS file.
Not Nullable Unique keys: At least one column from one of the unique keys must be not nullable. This is because NOALLOWNULLABLEKEYS is the default.

Note:
ALLOWNULLABLEKEYS is not valid for integrated Replicat.
If none of the preceding key types exist (even though there might be other types of keys defined on the table) Oracle GoldenGate constructs a pseudo key of all columns that the database allows to be used in a unique key, excluding virtual columns, UDTs, function-based columns, extended (32K) VARCHAR2/NVARCHAR2 columns, and any columns that are explicitly excluded from the Oracle GoldenGate configuration by an Oracle GoldenGate user.

Unless otherwise excluded due to the preceding restrictions, invisible columns are allowed in the pseudo key.

Note:

If there are other, non-usable keys on a table or if there are no keys at all on the table, Oracle GoldenGate logs an appropriate message to the report file. Constructing a key from all of the columns impedes the performance of Oracle GoldenGate on the source system. On the target, this key causes Replicat to use a larger, less efficient WHERE clause.

If a table does not have an appropriate key, or if you prefer the existing key(s) not to be used, you can define a substitute key if the table has columns that always contain unique values. You define this substitute key by including a KEYCOLS clause within the Extract TABLE parameter and the Replicat MAP parameter. The specified key will override any existing primary or unique key that Oracle GoldenGate finds. For more information, see Reference for Oracle GoldenGate.

Parent topic: Preparing the Database for Oracle GoldenGate

Configuring Logging Properties

Oracle GoldenGate relies on the redo logs to capture the data that it needs to replicate source transactions.

The Oracle redo logs on the source system must be configured properly before you start Oracle GoldenGate processing.

This section addresses the following logging levels that apply to Oracle GoldenGate. The logging level that you use depends on Oracle GoldenGate features that you are using.

Note:

Redo volume is increased as the result of this required logging. You can wait until you are ready to start Oracle GoldenGate processing to enable the logging.

This table shows the Oracle GoldenGate use cases for the different logging properties.

Logging option	Command	What it does	Use case
Forced logging mode	`ALTER DATABASE FORCE LOGGING`;	Forces the logging of all transactions and loads.	Strongly recommended for all Oracle GoldenGate use cases. `FORCE LOGGING` overrides any table-level `NOLOGGING` settings.
Minimum database-level supplemental logging	`ALTER DATABASE ADD SUPPLEMENTAL LOG DATA`	Enables minimal supplemental logging to add row-chaining information to the redo log.	Required for all Oracle GoldenGate use cases
Schema-level supplemental logging, default setting See Enabling Schema-level Supplemental Logging.	`ADD SCHEMATRANDATA`	Enables unconditional supplemental logging of the primary key and conditional supplemental logging of unique key(s) and foreign key(s) of all tables in a schema. All of these keys together are known as the scheduling columns.	Enables the logging for all current and future tables in the schema. If the primary key, unique key, and foreign key columns are not identical at both source and target, use `ALLCOLS`. Required when using DDL support.
Schema-level supplemental logging with unconditional logging for all supported columns. (See Enabling Schema-level Supplemental Logging for non-supported column types.)	`ADD SCHEMATRANDATA` with `ALLCOLS` option	Enables unconditional supplemental logging of all of the columns in a table, for all of the tables in a schema.	Used for bidirectional and active-active configurations where all column values are checked, not just the changed columns, when attempting to perform an update or delete. This takes more resources though allows for the highest level of real-time data validation and thus conflict detection. This method should also be used if they are going to be using the `HANDLECOLLISIONS` parameter for initial loads.
Schema-level supplemental logging, minimal setting	`ADD SCHEMATRANDATA` with `NOSCHEDULINGCOLS` option	Enables unconditional supplemental logging of the primary key and all valid unique indexes of all tables in a schema.	Use only for nonintegrated Replicat. This is the minimum required schema-level logging.
Table-level supplemental logging with built-in support for integrated Replicat See Enabling Table-level Supplemental Logging	`ADD TRANDATA`	Enables unconditional supplemental logging of the primary key and conditional supplemental logging of unique key(s) and foreign key(s) of a table. All of these keys together are known as the scheduling columns.	Required for all Oracle GoldenGate use cases unless schema-level supplemental logging is used. If the primary key, unique key, and foreign key columns are not identical at both source and target, use `ALLCOLS`.
Table-level supplemental logging with unconditional logging for all supported columns. (See Enabling Table-level Supplemental Logging for non-supported column types.)	`ADD TRANDATA` with `ALLCOLS` option	Enables unconditional supplemental logging of all of the columns of the table.	Used for bidirectional and active-active configurations where all column values are checked, not just the changed columns, when attempting to perform an update or delete. This takes more resources though allows for the highest level of real-time data validation and thus conflict detection. It can also be used when the source and target primary, unique, and foreign keys are not the same or are constantly changing between source and target.
Table-level supplemental logging, minimal setting	`ADD TRANDATA` with `NOSCHEDULINGCOLS` option	Enables unconditional supplemental logging of the primary key and all valid unique indexes of a table.	Use for nonintegrated Replicat and non-parallel Replicat. This is the minimum required table-level logging.

Note:

Oracle Databases must be in ARCHIVELOG mode so that Extract can process the log files.

Topics:

Parent topic: Preparing the Database for Oracle GoldenGate

Enabling Subset Database Replication Logging

Oracle strongly recommends putting the Oracle source database into forced logging mode. Forced logging mode forces the logging of all transactions and loads, overriding any user or storage settings to the contrary. This ensures that no source data in the Extract configuration gets missed.

There is a fine-granular database supplemental logging mode called Subset Database Replication available in LogMiner, which is the basic recommended mode for all Oracle GoldenGate and XStream clients. It replaces the previously used Minimum Supplemental Logging mode.

To know more, see ALTER DATABASEin the Oracle Database SQL Language Reference.

The subset database replication logging is enabled at CDB$ROOT (and all user-PDBs inherit it) currently.

Note:

Database-level primary key (PK) and unique index (UI) logging is only discouraged if you are replicating a subset of tables. You can use it with Live Standby, or if Oracle GoldenGate is going to replicate all tables, like to reduce the downtime for a migration or upgrade.

Perform the following steps to verify and enable, if necessary, subset database replication logging and forced logging.

Log in to SQL*Plus as a user with ALTER SYSTEM privilege.
Issue the following command to determine whether the database is in supplemental logging mode and in forced logging mode. If the result is YES for both queries, the database meets the Oracle GoldenGate requirement.
```
SELECT SUPPLEMENTAL_LOG_DATA_MIN, FORCE_LOGGING FROM V$DATABASE;
```
If the result is NO for either or both properties, continue with these steps to enable them as needed:
```
ALTER PLUGGABLE DATABASE pdbname ADD SUPPLEMENTAL LOG DATA SUBSET DATABASE REPLICATION;;
ALTER DATABASE FORCE LOGGING;
```
Issue the following command to verify that these properties are now enabled.
```
SELECT SUPPLEMENTAL_LOG_DATA_MIN, FORCE_LOGGING FROM V$DATABASE;
```
The output of the query must be YES for both properties.
Switch the log files.
```
ALTER SYSTEM SWITCH LOGFILE;
```

To perform dictionary validation, run the following command:

SELECT con_id, MINIMAL, SUBSET_REP, PRIMARY_KEY, UNIQUE_INDEX, FOREIGN_KEY, ALL_COLUMN FROM CDB_SUPPLEMENTAL_LOGGING;

The output of this query should be YES.

SUBSET_REP = YES

For the following query:

SELECT NAME, LOG_MODE, FORCE_LOGGING, SUPPLEMENTAL_LOG_DATA_MIN,
SUPPLEMENTAL_LOG_DATA_PK PK, SUPPLEMENTAL_LOG_DATA_UI UI,
SUPPLEMENTAL_LOG_DATA_FK FK, 
SUPPLEMENTAL_LOG_DATA_ALL,
SUPPLEMENTAL_LOG_DATA_SR FROM V$DATABASE;

For the query for SUPPLEMENTAL_LOG_DATA_SR the output should be YES and for SUPPLEMENTAL_LOG_DATA_MIN the output should be IMPLICIT.

To switch from earlier minimum supplemental logging to the new subset supplemental logging:

Drop the earlier higher levels on CDB$ROOT.

ALTER DATABASE DROP SUPPLEMENTAL LOG DATA (PRIMARY KEY) COLUMNS; 
ALTER DATABASE DROP SUPPLEMENTAL LOG DATA (UNIQUE) COLUMNS;
ALTER DATABASE DROP SUPPLEMENTAL LOG DATA;

Add only subset database replication mode:
```
ALTER PLUGGABLE DATABASE pdbname ADD SUPPLEMENTAL LOG DATA SUBSET DATABASE REPLICATION;
```
Ensure that all PDBs inherit this subset database replication mode.

Parent topic: Configuring Logging Properties

Enabling Schema-level Supplemental Logging

Oracle GoldenGate supports schema-level supplemental logging. Schema-level logging is required for an Oracle source database when using the Oracle GoldenGate DDL replication feature. In all other use cases, it is optional, but then you must use table-level logging instead (see Enabling Table-level Supplemental Logging).

By default, schema-level logging automatically enables unconditional supplemental logging of the primary key and conditional supplemental logging of unique key(s) and foreign key(s) of all tables in a schema. Options enable you to alter the logging as needed.

Note:

Oracle strongly recommends using schema-level logging rather than table-level logging, because it ensures that any new tables added to a schema are captured if they satisfy wildcard specifications. This method is also recommended because any changes to key columns are automatically reflected in the supplemental log data too. For example, if a key changes, there is no need to issue ADD TRANDATA.

Perform the following steps on the source system to enable schema-level supplemental logging.

Start the command line on the source system.
Issue the DBLOGIN command with the alias of a user in the credential store who has privilege to enable schema-level supplemental logging.
```
DBLOGIN USERIDALIAS alias
```
See USERIDALIAS in Reference for Oracle GoldenGate for more information about USERIDALIAS and additional options.
When using ADD SCHEMATRANDATA or ADD TRANDATA on a multitenant database, you can either log directly into the PDB and perform the command. Alternately, if you are logging in at the root level (using a C## user), then you must include the PDB. Issue the ADD SCHEMATRANDATA command for each schema for which you want to capture data changes with Oracle GoldenGate.
```
ADD SCHEMATRANDATA pdb.schema [ALLCOLS | NOSCHEDULINGCOLS]
```
Where:
- Without options, ADD SCHEMATRANDATA schema enables the unconditional supplemental logging on the source system of the primary key and the conditional supplemental logging of all unique key(s) and foreign key(s) of all current and future tables in the given schema. Unconditional logging forces the primary key values to the log whether or not the key was changed in the current operation. Conditional logging logs all of the column values of a foreign or unique key if at least one of them was changed in the current operation. The default is optional to support nonintegrated Replicat but is required to support integrated Replicat because primary key, unique keys, and foreign keys must all be available to the inbound server to compute dependencies. For more information about integrated Replicat, see Deciding Which Apply Method to Use.
- ALLCOLS can be used to enable the unconditional supplemental logging of all of the columns of a table and applies to all current and future tables in the given schema. Use to support integrated Replicat when the source and target tables have different scheduling columns. (Scheduling columns are the primary key, the unique key, and the foreign key.)
- NOSCHEDULINGCOLS logs only the values of the primary key and all valid unique indexes for existing tables in the schema and new tables added later. This is the minimal required level of schema-level logging and is valid only for Replicat in nonintegrated mode.
In the following example, the command enables default supplemental logging for the hr schema.
```
ADD SCHEMATRANDATA pdbeast.hr ALLCOLS
```
In the following example, the command enables the supplemental logging only for the primary key and valid unique indexes for the HR schema.
```
ADD SCHEMATRANDATA pdbeast.hr NOSCHEDULINGCOLS
```

Parent topic: Configuring Logging Properties

Enabling Table-level Supplemental Logging

Enable table-level supplemental logging on the source system in the following cases:

To enable the required level of logging when not using schema-level logging (see Enabling Schema-level Supplemental Logging). Either schema-level or table-level logging must be used. By default, table-level logging automatically enables unconditional supplemental logging of the primary key and conditional supplemental logging of unique key(s) and foreign key(s) of a table. Options enable you to alter the logging as needed.
To prevent the logging of the primary key for any given table.
To log non-key column values at the table level to support specific Oracle GoldenGate features, such as filtering and conflict detection and resolution logic.
If the key columns change on a table that only has table-level supplemental logging, you must perform ADD TRANDATA on the table prior to allowing any DML activity on the table.

Perform the following steps on the source system to enable table-level supplemental logging or use the optional features of the command.

Run the command line on the source system.
Issue the DBLOGIN command using the alias of a user in the credential store who has privilege to enable table-level supplemental logging.
```
DBLOGIN USERIDALIAS alias
```
See USERIDALIAS in Reference for Oracle GoldenGatefor more information about DBLOGIN and additional options.
Issue the ADD TRANDATA command.
```
ADD TRANDATA [PDB.]schema.table [, COLS (columns)] [, NOKEY] [, ALLCOLS | NOSCHEDULINGCOLS]
```
Where:
- PDB is the name of the root container or pluggable database if the table is in a multitenant container database.
- schema is the source schema that contains the table.
- table is the name of the table. See Specifying Object Names in Oracle GoldenGate Input in Administering Oracle GoldenGate for instructions for specifying object names.
- ADD TRANDATA without other options automatically enables unconditional supplemental logging of the primary key and conditional supplemental logging of unique key(s) and foreign key(s) of the table. Unconditional logging forces the primary key values to the log whether or not the key was changed in the current operation. Conditional logging logs all of the column values of a foreign or unique key if at least one of them was changed in the current operation. The default is optional to support nonintegrated Replicat (see also NOSCHEDULINGCOLS) but is required to support integrated Replicat because primary key, unique keys, and foreign keys must all be available to the inbound server to compute dependencies. For more information about integrated Replicat, see Deciding Which Apply Method to Use.
- ALLCOLS enables the unconditional supplemental logging of all of the columns of the table. Use to support integrated Replicat when the source and target tables have different scheduling columns. (Scheduling columns are the primary key, the unique key, and the foreign key.)
- NOSCHEDULINGCOLS is valid for Replicat in nonintegrated mode only. It issues an ALTER TABLE command with an ADD SUPPLEMENTAL LOG DATA ALWAYS clause that is appropriate for the type of unique constraint that is defined for the table, or all columns in the absence of a unique constraint. This command satisfies the basic table-level logging requirements of Oracle GoldenGate when schema-level logging will not be used. See Ensuring Row Uniqueness in Source and Target Tables for how Oracle GoldenGate selects a key or index.
- COLS columns logs non-key columns that are required for a KEYCOLS clause or for filtering and manipulation. The parentheses are required. These columns will be logged in addition to the primary key unless the NOKEY option is also present.
- NOKEY prevents the logging of the primary key or unique key. Requires a KEYCOLS clause in the TABLE and MAP parameters and a COLS clause in the ADD TRANDATA command to log the alternate KEYCOLS columns.
If using ADD TRANDATA with the COLS option, create a unique index for those columns on the target to optimize row retrieval. If you are logging those columns as a substitute key for a KEYCOLS clause, make a note to add the KEYCOLS clause to the TABLE and MAP statements when you configure the Oracle GoldenGate processes.

Parent topic: Configuring Logging Properties

Using the Extract Automated Capture Mode

The automated capture (auto capture) mode allows automatically capturing the tables that have been enabled for Oracle GoldenGate auto capture.

See How to Capture Supplemental Logging for Oracle GoldenGate in the Oracle Database Utilities guide.

Here are some benefits of using the auto capture mode:

Easy to configure captured table set
No requirement to update TABLE/TABLEEXCLUDE parameter
No need to stop or restart Extract when captured table set changes

Enabling Extract Auto Capture Mode

To enable the Extract Auto Capture Mode, enable the auto capture mode using TRANLOGOPTIONS:

TRANLOGOPTIONS ENABLE_AUTO_CAPTURE | DISABLE_AUTO_CAPTURE

When Extract is running in the auto capture mode, don't filter an LCR if the object is not part of exclusion list set by TABLE EXCLUDE parameter or any inclusion list set by TABLE parameter.

The LIST TABLES command shows the list of tables enabled for AUTO_CAPTURE.

See DML Auto Capture and Supported Objects and Operations in Oracle DDL to know about the DML and DDL considerations.

Also see this article Oracle GoldenGate 21c: Auto Capture of Tables to learn more.

Parent topic: Preparing the Database for Oracle GoldenGate

Additional Oracle GoldenGate Configuration for Your Database

This chapter contains additional configuration considerations that may apply to your database environment.

Topics:

Installing Support for Oracle Sequences
To support Oracle sequences, you must install some database procedures.
Handling Other Database Properties
This topic describes the database properties that may affect Oracle GoldenGate and the parameters that you can use to resolve or work around the condition.

Parent topic: Preparing the Database for Oracle GoldenGate

Installing Support for Oracle Sequences

To support Oracle sequences, you must install some database procedures.

To Install Oracle Sequence Objects

In SQL*Plus, connect to the source and target Oracle systems as SYSDBA.
If you already assigned a database user to support the Oracle GoldenGate DDL replication feature, you can skip this step. Otherwise, in SQL*Plus on both systems create a database user that can also be the DDL user.
```
CREATE USER DDLuser IDENTIFIED BY password; 
GRANT CONNECT, RESOURCE, DBA TO DDLuser;
```
From the Oracle GoldenGate installation directory on each system, run GGSCI.
In GGSCI, issue the following command on each system.
```
EDIT PARAMS ./GLOBALS
```
In each GLOBALS file, enter the GGSCHEMA parameter and specify the schema of the DDL user that you created earlier in this procedure.
```
GGSCHEMA schema
```
Save and close the files.
In SQL*Plus on the source system, issue the following statement in SQL*Plus.
```
ALTER TABLE sys.seq$ ADD SUPPLEMENTAL LOG DATA (PRIMARY KEY) COLUMNS;
```

To capture the sequence from a multitenant database

Create an Oracle GoldenGate user in each PDB that you need to capture sequences from.
Add the user to the GLOBALS parameter file. It is easier if you use the same user for each PDB, if you don't then you need to change the GLOBALS file each time you do step 3.
Log into Admin Client or GGSCI.
Connect to the root container on the source using DBLOGIN.
Issue the FLUSH SEQUENCE command for each PDB.

If you don't want to keep these database accounts, you can drop the user or deactivate the account.

Here is an example of the entire process:

Environment information
            OGG 19.1 Oracle 12c to Oracle 12c Replication, Integrated 
            Extract, Parallel Replicat
            Source: CDB GOLD, PDB CERTMISSN 
            Target: CDB PLAT, PDB CERTDSQ 
            Source Oracle GoldenGate Configuration
            Container User: C##GGADMIN 
            PDB User for Sequences: GGATE

When prompted, enter GGATE

GLOBALS 
            GGSCHEMA GGATE 
         Flush Sequence 
            GGSCI> DBLOGIN USERIDALIAS GGADMIN DOMAIN GOLD_QC_CDB$ROOT 
            GGSCI> FLUSH SEQUENCE CERTMISSN.SRCSCHEMA1.
Target OGG Configuration 
         PDB User: GGATE 
         Run @sequence 
                sqlplus / as sysdba 
                SQL> alter session set container=CERTDSQ;
                SQL> @sequence

When prompted enter GGATE.

Replicating sequences is required for High Availability (HA) and DR scenarios:

For migrations, you need rebuild the sequences on the target during the switchover, or increase them to a higher value just prior to the switchover.
Make sure you place the sequences into their own Replicat.

Parent topic: Additional Oracle GoldenGate Configuration for Your Database

Handling Other Database Properties

This topic describes the database properties that may affect Oracle GoldenGate and the parameters that you can use to resolve or work around the condition.

The following table lists the database properties and the associated concern/resolution.

Database Property	Concern/Resolution
Table with interval partitioning	To support tables with interval partitioning, make certain that the `WILDCARDRESOLVE` parameter remains at its default of `DYNAMIC`.
Table with virtual columns	Virtual columns are not logged, and Oracle does not permit DML on virtual columns. You can, however, capture this data and map it to a target column that is not a virtual column by doing the following: Include the table in the Extract `TABLE` statement and use the `FETCHCOLS` option of `TABLE` to fetch the value from the virtual column in the database. In the Replicat `MAP` statement, map the source virtual column to the non-virtual target column.
Table with inherently updateable view	To replicate to an inherently updateable view, define a key on the unique columns in the updateable view by using a `KEYCOLS` clause in the same `MAP` statement in which the associated source and target tables are mapped.
Redo logs or archives in different locations	The `TRANLOGOPTIONS` parameter contains options to handle environments where the redo logs or archives are stored in a different location than the database default or on a different platform from that on which Extract is running. For more information, see Reference for Oracle GoldenGate.
`TRUNCATE` operations	To replicate `TRUNCATE` operations, choose one of two options: Standalone `TRUNCATE` support by means of the `GETTRUNCATES` parameter replicates `TRUNCATE TABLE`, but no other `TRUNCATE` options. Use only if not using Oracle GoldenGate DDL support. The full DDL support replicates `TRUNCATE TABLE`, `ALTER TABLE TRUNCATE` `PARTITION`, and other DDL.
Sequences	To replicate DDL for sequences (`CREATE`, `ALTER`, `DROP`, `RENAME`), use Oracle GoldenGate DDL support. To replicate just sequence values, use the `SEQUENCE` parameter in the Extract parameter file. This does not require the Oracle GoldenGate DDL support environment. For more information, see Reference for Oracle GoldenGate.

Parent topic: Additional Oracle GoldenGate Configuration for Your Database