3.171 TABLE | MAP

Valid For

TABLE is valid for Extract. MAP is valid for Replicat

Description

The TABLE and MAP parameters control the selection, mapping, and manipulation of the objects that are to be affected by an Oracle GoldenGate process. These parameters work as follows:

  • Use the TABLE parameter in an Extract parameter file to specify one or more objects that are to be captured from the data source by the Extract process. TABLE options specify processing work such as filtering and token definitions that must be performed before Extract writes the captured data to the Oracle GoldenGate trail.

  • Use the MAP parameter in the Replicat parameter file to map the data from the source objects to the appropriate target objects. MAP options specify processing work such as filtering, conversion, and error handling that must be performed before the data is applied to the target objects. Each target object that you want to synchronize with a source object must be associated with that source object by means of a MAP parameter. Multiple source-target relationships can be specified by means of a wildcard.

TABLE and MAP are valid for initial load configurations and for online processes configured to support the replication of transactional changes.

You can process the following objects with TABLE and MAP:

  • Indexes

  • Triggers

  • Materialized views

  • Tables

To specify a sequence for capture by Extract, use the SEQUENCE parameter.

Note:

Oracle GoldenGate supports the replication of the actual data values of Oracle materialized views. Oracle GoldenGate supports the replication of Oracle and Teradata DDL for indexes and triggers, but not the content of those objects.

You can use one or more TABLE or MAP statements in a parameter file, with or without wildcards, to specify all of the objects that you want to process.

You can exclude objects from a wildcarded TABLE or MAP statement with the TABLEEXCLUDE and MAPEXCLUDE parameters. Additional exclusion parameters are CATALOGEXCLUDE, SCHEMAEXCLUDE, and EXCLUDEWILDCARDOBJECTSONLY.

For more information about using TABLE and MAP, see Administering Oracle GoldenGate for Windows and UNIX.

Default

None

Syntax for TABLE

For tables, you can use all of the TABLE options. For non-table objects, use TABLE only to specify an object for capture.

TABLE source_table[, TARGET target_table]
[, ATTRCHARSET (charset)]
[, CHARSET character_set]
[, COLCHARSET character_set]
[, COLMAP (column_mapping)]
[, {COLS | COLSEXCEPT} (column_list)]
[, {DEF | TARGETDEF} template]
[, EVENTACTIONS action]
[, EXITPARAM 'parameter']
[, {FETCHCOLS | FETCHCOLSEXCEPT} (column_list)]
[, {FETCHMODCOLS | FETCHMODCOLSEXCEPT} (column_list)]
[, FETCHBEFOREFILTER]
[, FILTER (filter_clause)]
[, GETBEFORECOLS (column_specification)]
[, KEYCOLS (columns)]
[, SQLEXEC (SQL_specification)]
[, SQLPREDICATE 'WHERE where_clause']
[, TOKENS (token_definition)]
[, TRIMSPACES | NOTRIMSPACES]
[, TRIMVARSPACES | NOTRIMVARSPACES]
[, WHERE (clause)]
;

Syntax for MAP

MAP source_table, TARGET target_table
[, COLMAP (column_mapping)]
[, COMPARECOLS (column_specification)]
[, COORDINATED]
[, {DEF | TARGETDEF} template]
[, EXCEPTIONSONLY]
[, EXITPARAM 'parameter']
[, EVENTACTIONS (action)]
[, FILTER (filter_clause)]
[, HANDLECOLLISIONS | NOHANDLECOLLISIONS]
[, INSERTALLRECORDS]
[, INSERTAPPEND | NOINSERTAPPEND]
[, KEYCOLS (columns)]
[, MAPEXCEPTION (exceptions_mapping)]
[, MAPINVISIBLECOLUMNS | NOMAPINVISIBLECOLUMNS]
[, REPERROR (error, response)]
[, RESOLVECONFLICT (conflict_resolution_specification)]
[, SQLEXEC (SQL_specification)]
[, THREAD (thread_ID)]
[, THREADRANGE (thread_range[, column_list])]
[, TRIMSPACES | NOTRIMSPACES]
[, TRIMVARSPACES | NOTRIMVARSPACES]
[, WHERE (clause)]
;

TABLE and MAP Options

The following table summarizes the options that are available for the TABLE and MAP parameters. Note that not all options are valid for both parameters.

Table 3-34 Summary of TABLE and MAP Syntax Components

Component Description Valid For
TABLE source_table[, TARGET taget_table]

Specifies the source object in a TABLE statement for Extract and an optional mapping to a target object. Use in the Extract parameter file.

TABLE

MAP source_table, TARGET target_table

Specifies the source-target object mapping for the Replicat process. Use in the Replicat parameter file.

MAP

ATTRCHARSET (charset)

specifies the source character set information at UDT attribute level.

TABLE

CHARSET character_set

Specifies any supported character set.

TABLE

COLCHARSET character_set

Specifies any supported character set.

TABLE

COLMAP (column_mapping)

Maps records between different source and target columns.

TABLE and MAP

{COLS | COLSEXCEPT} (column_list)

Selects or excludes columns for processing.

TABLE

COMPARECOLS (column_specification)

Specifies columns to use for conflict detection and resolution.

TABLE and MAP

COORDINATED

Forces a transaction to be processed as a barrier transaction.

MAP

{DEF| TARGETDEF} template

Specifies a source-definitions or target-definitions template.

TABLE and MAP

EXCEPTIONSONLY

Specifies that the MAP statement is an exceptions MAP statement.

MAP

EVENTACTIONS (action)

Triggers an action based on a record that satisfies a specified filter rule.

TABLE and MAP

EXITPARAM 'parameter'

Passes a parameter in the form of a literal string to a user exit.

TABLE and MAP

FETCHBEFOREFILTER

Directs the FETCHCOLS or FETCHCOLSEXCEPT action to be performed before a filter is executed.

TABLE

{FETCHCOLS | FETCHCOLSEXCEPT} (column_list)

Enables the fetching of column values from the source database when the values are not in the transaction record.

TABLE

{FETCHMODCOLS | FETCHMODCOLSEXCEPT} (column_list)

Forces column values to be fetched from the database when the columns are present in the transaction log.

TABLE

FILTER (filter_clause)

Selects records based on a numeric value. FILTER provides more flexibility than WHERE.

TABLE and MAP

GETBEFORECOLS (column_specification)

Forces before images of columns to be captured and written to the trail.

TABLE

HANDLECOLLISIONS | NOHANDLECOLLISIONS

Reconciles the results of changes made to the target table by an initial load process with those applied by a change-synchronization group.

MAP

INSERTALLRECORDS

Applies all row changes as inserts.

MAP

INSERTAPPEND | NOINSERTAPPEND

Controls whether or not Replicat uses an Oracle APPEND hint for INSERT statements.

MAP

KEYCOLS (columns)

Designates columns that uniquely identify rows.

TABLE and MAP

MAPEXCEPTION (exceptions_mapping)

Specifies that the MAP statement contains exceptions handling for wildcarded tables.

MAP

MAPINVISIBLECOLUMNS | NOMAPINVISIBLECOLUMNS

Controls whether or not Replicat includes invisible columns in Oracle target tables for default column mapping. For invisible columns in Oracle target tables that use explicit column mapping, they are always mapped so do not require this option.

MAP

REPERROR (error, response)

Controls how Replicat responds to errors when executing the MAP statement.

MAP

RESOLVECONFLICT (conflict_resolution_specification)

Specifies rules for conflict resolution.

MAP

SQLEXEC (SQL_specification)

Executes stored procedures and queries.

TABLE and MAP

SQLPREDICATE 'WHERE where_clause'

Enables a WHERE clause to select rows for an initial load.

TABLE

THREAD (thread_ID)

Valid for Replicat in coordinated mode. Specifies that the MAP statement will be processed by the specified Replicat thread.

MAP

THREADRANGE (thread_range, column_list)

Valid for Replicat in coordinated mode. Specifies that the MAP statement will be processed by the specified range of Replicat threads.

MAP

TOKENS (token_definition)

Defines user tokens.

TABLE

TRIMSPACES | NOTRIMSPACES

Controls whether trailing spaces are trimmed or not when mapping CHAR to VARCHAR columns.

TABLE and MAP

TRIMVARSPACES | NOTRIMVARSPACES

Controls whether trailing spaces are trimmed or not when mapping VARCHAR to CHAR or VARCHAR columns.

TABLE and MAP

WHERE (clause)

Selects records based on conditional operators.

TABLE and MAP

;

(Semicolon) Terminates the TABLE or MAP statement and is required.

TABLE and MAP

TABLE source_table[, TARGET taget_table]

TABLE is valid in an Extract parameter file.

Use TABLE to specify a source object for which you want Extract to capture data. Specify the fully qualified two-part or three-part name of the object, such as schema.table or catalog.schema.table. You can use a wildcard to specify multiple objects with one TABLE statement. To specify object names and wildcards correctly, see Administering Oracle GoldenGate for Windows and UNIX.

Use the TARGET option only when Extract must refer to a target definitions file (specified with the TARGETDEFS parameter) to perform conversions or when the COLMAP option is used to map columns. Otherwise, it can be omitted from a TABLE parameter. Column mapping with COLMAP and conversion work usually are performed on the target system to minimize the impact of replication activities on the source system, but can be performed on the source system if required. For example, column mapping and conversion can be performed on the source system in a configuration where there are multiple sources and one target. In this scenario, it may be easier to manage one target definitions file rather than managing a definitions file for each source database, especially if there are frequent application changes that require new definitions files to be generated.

Using TARGET in a TABLE parameter identifies the metadata of the extracted data based on the target structure, rather than that of the source, to reflect the structure of the record that is reflected in the definitions file or the column map. Do not use three-part names if TARGET specifies tables in a target Oracle container database or SQL/MX database. Replicat can only connect to one container or catalog, so it is assumed that the container or catalog portion of the name is the same as the one that Replicat logs into (as specified with USERID, USERIDALIAS, or TARGETDB, depending on the database).

If no other TABLE syntax options are required to process the specified source data, you can use a simple TABLE statement, making sure to terminate it with a semicolon.

TABLE sales.customers;

The following shows the use of a wildcard to specify multiple tables:

TABLE sales.*;

The preceding TABLE statements direct Extract to capture all supported column data for the specified objects and write it to the trail without performing any filtering, conversion, or other manipulation.

MAP source_table, TARGET target_table

MAP is valid in a Replicat parameter file.

Use MAP to specify a source object, and use TARGET to specify the target object to which Replicat applies the replicated source data. Together, the MAP and TARGET clause comprise a mapping.

  • For MAP source_table, specify the source object. Specify the fully qualified two-part or three-part name of the object, such as schema.table or catalog.schema.table. You can use a wildcard to specify multiple source objects.

  • For TARGET target_table, specify a two-part name, even if the target is a container database or SQL/MX database. Replicat can only connect to one container or catalog, so it is assumed that the container or catalog portion of the name is the same as the one that Replicat logs into (as specified with USERID, USERIDALIAS, or TARGETDB, depending on the database). You can use a wildcard to specify multiple target objects.

The following shows the use of a wildcard to specify multiple tables. Note that the TARGET clause does not include the tab prefix before the wildcard. That specification would be invalid, because the wildcard would be resolved as sales.tabtab1, sales.tabtab2, and so forth.

MAP sales.tab*, TARGET sales.*;

To specify object names and wildcards correctly in the MAP and TARGET clauses, see Administering Oracle GoldenGate for Windows and UNIX.

If no filtering, mapping, or other work is required for the objects, you can use simple MAP statements like the following, making sure to terminate each one with a semicolon.

MAP sales.customers, TARGET sales.customers;
MAP fin.*, TARGET fin.*;

ATTRCHARSET (charset)

ATTRCHARSET is valid for TABLE.

Use the ATTRCHARSET clause to specify the source character set information at UDT attribute level. It overrides the character set defined in the trail file or specified by SOURCECHARSET, CHARSET, or COLCHARSET parameters.

Valid values are character set names and valid UDT attribute names. Wildcard attribute names are supported. For example:

TABLE SCHEMA.T*,
  ATTRCHARSET(WE8DEC, col*.attr1, col1.attr*.attr3);

CHARSET character_set

CHARSET is valid for TABLE.

Use the CHARSET clause to specify any supported character set. See CHARSET for more information.

COLCHARSET character_set

COLCHARSET is valid for TABLE.

Use the COLCHARSET clause to specify any supported character set. See COLCHARSET for more information.

COLMAP (column_mapping)

COLMAP is valid for TABLE and MAP.

Use COLMAP to:

  • Map individual source columns to target columns when the source and target columns have different names.

  • Specify default column mapping when the source and target names are identical.

COLMAP provides instructions for selecting, translating, and moving column data from a source column to a target column.

Note:

To create global rules for column mapping across all tables in subsequent MAP statements, use the COLMATCH parameter.

Getting More Information About Configuring Column Mapping

See Administering Oracle GoldenGate for Windows and UNIX for more information about using COLMAP. To use COLMAP, related configuration considerations must be taken into account, such as whether source and target column structures are identical or different and whether global column mapping parameters may be sufficient.

Syntax

COLMAP (
[USEDEFAULTS, ]
target_column = source_expression [BINARYINPUT]
[, ...]
)
USEDEFAULTS

Automatically maps source and target columns that have the same name if they were not specified in an explicit column mapping. The data types are translated automatically, as needed, based on the local data-definitions file. USEDEFAULTS eliminates the need for an explicit column mapping if those columns have the same name and the data does not require any filtering or conversion.

Specify USEDEFAULTS before explicit column mappings in the COLMAP clause. For additional information about default column mapping in COLMAP, see Administering Oracle GoldenGate for Windows and UNIX.

target_column = source_expression

Defines an explicit source-target column mapping.

target_column

Specifies the name of the target column. For supported characters in column names, see Administering Oracle GoldenGate for Windows and UNIX.

source_expression

Can be any of the following:

  • The name of a source column, such as ORD_DATE

  • A numeric constant, such as 123

  • A string constant within single quotes, such as 'ABCD'

  • An expression using an Oracle GoldenGate column-conversion function, such as @STREXT (COL1, 1, 3). See "Column Conversion Functions" for more information.

BINARYINPUT

Use BINARYINPUT when the target column is defined as a binary data type, such as RAW or BLOB, but the source input contains binary zeros in the middle of the data. Use BINARYINPUT when replicating a full Enscribe record defined as a single column into a target column. The source input is handled as binary input, and replacement of data values is suppressed.

Example 1   
MAP ggs.tran, TARGET ggs.tran2, COLMAP (loc2 = loc, type2 = type);
Example 2   
TABLE ggs.tran, COLMAP (SECTION = @STRCAT('\u00a7', SECTION ));

{COLS | COLSEXCEPT} (column_list)

COLS and COLSEXCEPT are valid for TABLE.

Use COLS and COLSEXCEPT to control the columns for which data is captured.

  • COLS specifies columns that contain the data that you want to capture. When COLS is used, all columns that are not in the COLS list are ignored by Oracle GoldenGate.

  • COLSEXCEPT specifies columns to exclude from being captured. When COLSEXCEPT is used, all columns that are not in the COLSEXCEPT list are captured by Oracle GoldenGate. For tables with numerous columns, COLSEXCEPT may be more efficient than listing each column with COLS.

    Caution:

    Do not exclude key columns, and do not use COLSEXCEPT to exclude columns that contain data types that are not supported by Oracle GoldenGate. COLSEXCEPT does not exclude unsupported data types.

To use COLS, the following is required:

  • The table must have one or more key columns, or a substitute key must be defined with the KEYCOLS option. See "KEYCOLS (columns)".

  • The key columns or the columns specified with KEYCOLS must be included in the column list that is specified with COLS. Otherwise, they will not be captured, and an error will be generated during processing.

Without a primary key, a unique key, or a KEYCOLS clause in the TABLE statement, Oracle GoldenGate uses all of the columns in the table, rendering COLS unnecessary.

Note:

Do not use this option for tables that are processed in pass-through mode by a data-pump Extract group.

Syntax

{COLS | COLSEXCEPT} (column [, ...])
column

The name of a column. To specify multiple columns, create a comma-delimited list, for example:

COLS (name, city, state, phone)

Note:

If the database only logs values for columns that were changed in an update operation, a column specified for capture with COLS might not be available. To make those columns available, use the FETCHCOLS option in the TABLE statement or enable supplemental logging for the column.

Example

The COLS clause in this example captures only columns 1 and 3, whereas the COLSEXCEPT clause captures all columns except columns 1 and 3.

TABLE hq.acct, COLS (col1, col3);
TABLE hq.sales, COLSEXCEPT (col1, col3);

COMPARECOLS (column_specification)

COMPARECOLS is valid for MAP.

Use COMPARECOLS to specify the columns that Replicat uses to detect and resolve update or delete conflicts when configured with the RESOLVECONFLICT option of MAP in a multi-master configuration. A conflict is a mismatch between the before image of a record in the trail and the current data in the target table.

To use COMPARECOLS, the before image must be available in the trail record by means of the GETBEFORECOLS parameter in the Extract TABLE statement. The specified columns must exist in the target database and also be part of the Replicat configuration (satisfy the TARGET specification with or without a COLMAP clause).

Only scalar data types are supported by COMPARECOLS as comparison columns. A scalar data type can be used in a WHERE clause, has a single, atomic value and no internal components. Scalar data types supported by Oracle GoldenGate include the following, but not LOBs.

  • Numeric data types

  • Date data types

  • Character data types

Some examples of non-scalar data types are spatial data, user-defined data types, large objects (LOB), XML, reference data types, and RAW. A row being considered for CDR can include non-scalar data so long as the conflict is not in the non-scalar data itself.

To specify conflict resolution routines, use the RESOLVECONFLICT option of MAP. COMPARECOLS and RESOLVECONFLICT can be in any order in the MAP statement.

Getting More Information About Configuring the CDR Feature

See Administering Oracle GoldenGate for Windows and UNIX for more information about configuring conflict detection and resolution.

Syntax

COMPARECOLS(
{ON UPDATE | ON DELETE}
{ALL | KEY | KEYINCLUDING (col[,...]) | ALLEXCLUDING (col[,...]) }
[,...]
)
{ON UPDATE | ON DELETE}

Specifies whether the before image of the specified columns should be compared for updates or deletes. You can use ON UPDATE only, ON DELETE only, or both. If using both, specify them within the same COMPARECOLS clause. See the example for how to use both.

{ALL | KEY | KEYINCLUDING (col[,...]) | ALLEXCLUDING (col[,...])}

Specifies the columns for which a before image is captured.

ALL

Compares using all columns in the target table. An error is generated if any corresponding before images are not available in the trail. Using ALL imposes the highest processing load for Replicat, but allows conflict-detection comparisons to be performed using all columns for maximum accuracy.

KEY

Compares only the primary key columns. This is the fastest option, but does not permit the most accurate conflict detection, because keys can match but non-key columns could be different.

KEYINCLUDING

Compares the primary key columns and the specified column or columns. This is a reasonable compromise between speed and detection accuracy.

ALLEXCLUDING

Compares all columns except the specified columns. For tables with numerous columns, ALLEXCLUDING may be more efficient than KEYINCLUDING. Do not exclude key columns.

Example 1   

In the following example, the key columns plus the name, address, and salary columns are compared for conflicts.

MAP src, TARGET tgt
COMPARECOLS (
ON UPDATE KEYINCLUDING (name, address, salary),
ON DELETE KEYINCLUDING (name, address, salary));
Example 2   

In the following example, the comment column is ignored and all other columns are compared for conflicts.

MAP src, TARGET tgt
COMPARECOLS (ON UPDATE ALLEXCLUDING (comment))

COORDINATED

COORDINATED is valid for MAP. This option is valid when Replicat is in coordinated mode.

Use the COORDINATED option to force transactions made on objects in the same MAP statement to be processed as barrier transactions. It causes all of the threads across all MAP statements to synchronize to the same trail location. The synchronized position is the beginning of the transaction that contains a record that satisfies a MAP that contains the COORDINATED keyword. The transaction is then applied atomically by a single thread, which is either the thread with the lowest thread ID among the currently running threads or a dedicated thread with the ID of 0 if USEDEDICATEDCOORDINATIONTHREAD is specified in the parameter file.

THREAD and THREADRANGE clauses specified in conjunction with COORDINATED are ignored because the record will not be applied by the designated thread(s). The COORDINATED keyword results in temporarily suspending parallelism so that the target tables are in a consistent state before the force-coordinated transaction is applied. After this point, parallel execution commences again.

Replicat by default coordinates transactions in which the primary key is updated, transactions that perform DDL, and certain EVENTACTIONS actions. COORDINATED provides for explicit coordination.

See Administering Oracle GoldenGate for Windows and UNIX for more information about coordinated Replicat.

Syntax

COORDINATED

Example

The following is an example of the use of the COORDINATED option. In this example, business rules require that the target tables be in a consistent state before Replicat executes transactions that include SQLEXEC operations on the objects specified in the MAP statement. Parallelism must be temporarily converted to serial SQL processing in this case.

Given the following MAP statement, if another thread inserts into t2 a record with a value of 100 for col_val before the insert to t1 is performed by thread 1, then the SQLEXEC will delete the row. If other threads are still processing the record that has the value of 100, the SQLEXEC fails. The results of this MAP statement are, therefore, not predictable.

MAP u1.t1, TARGET u2.t1 SQLEXEC (ID test2, QUERY ' delete from u2.t2 where col_val =100 ', NOPARAMS)), THREAD(1);

Conversely, when COORDINATED is used, all of the threads synchronize at a common point, including the one processing the col_val=100 record, thereby removing the ambiguity of the results.

MAP u1.t1, TARGET u2.t1 SQLEXEC (ID test2, QUERY ' delete from u2.t2 where col_val =100 ', NOPARAMS)), THREAD(1), COORDINATED;

{DEF| TARGETDEF} template

DEF and TARGETDEF are valid for TABLE and MAP.

Use DEF and TARGETDEF to specify the name of a definitions template that was created by the DEFGEN utility.

  • DEF specifies a source-definitions template.

  • TARGETDEF specifies a target-definitions template.

A template is based on the definitions of a specific table. It enables new tables that have the same definitions as the original table to be added to the Oracle GoldenGate configuration without running DEFGEN for them, and without having to stop and start the Oracle GoldenGate process. The definitions in the template are used for definitions lookups.

Getting More Information About Creating Definitions Templates

For more information about templates and DEFGEN, see Administering Oracle GoldenGate for Windows and UNIX.

Syntax

{DEF | TARGETDEF} template
template

The name of one of the following definitions templates generated by the DEFGEN utility:

  • Use DEF to specify a source-definitions template generated by the DEF option of the TABLE parameter in the DEFGEN parameter file.

  • Use TARGETDEF to specify a target-definitions template generated by the TARGETDEF option of the TABLE parameter in the DEFGEN parameter file.

The definitions contained in the template must be identical to the definitions of the table or tables that are specified in the same TABLE or MAP statement.

Case-sensitivity of the template name is observed when the name is specified the same way that it is stored in the database. Make certain that the template name is specified the same way in both the DEF or TARGETDEF clause in this TABLE or MAP statement, and in the DEFGEN parameter file that created the template.

Example 1   

This example shows a case-insensitive template name.

MAP acct.cust*, TARGET acct.cust*, DEF custdef;
Example 2   

This example shows a case-sensitive template name when the database requires quotes to enforce case-sensitivity.

TABLE acct.cust*, DEF "CustDef";
Example 3   

This example shows a case where both DEF and TARGETDEF are used.

MAP acct.cust*, TARGET acc.cust*, DEF custdef, TARGETDEF tcustdef;

EXCEPTIONSONLY

EXCEPTIONSONLY is valid for MAP.

Use EXCEPTIONSONLY in an exceptions MAP statement intended for error handling. The exceptions MAP statement must follow the MAP statement for which errors are anticipated. The exceptions MAP statement executes only if an error occurs for the last record processed in the preceding regular MAP statement.

To use EXCEPTIONSONLY, use a REPERROR statement with the EXCEPTION option either within the regular MAP statement or at the root of the parameter file. See "REPERROR" for more information.

Note:

If using the Oracle GoldenGate Conflict Detection and Resolution (CDR) feature, a REPERROR with EXCEPTION is not needed. CDR automatically sends all operations that cause errors to the exceptions MAP statement.

The exceptions MAP statement must specify the same source table as in the regular MAP statement, but the target table in the exceptions MAP statement must be an exceptions table.

Note:

See "MAPEXCEPTION (exceptions_mapping)" to support wildcarded object names.

Getting More Information About Configuring Exceptions Handling

For more information about configuring exceptions handling with an exceptions MAP statement, see Administering Oracle GoldenGate for Windows and UNIX.

Syntax

EXCEPTIONSONLY

EVENTACTIONS (action)

EVENTACTIONS is valid for TABLE and MAP. Some options apply only to one or the other parameter and are noted as such in the descriptions.

Use EVENTACTIONS to cause the process to take a defined action based on a record in the trail, known as the event record, that qualifies for a specific filter rule. You can use this system, known as the event marker system (or event marker infrastructure) to customize processing based on database events. For example, you can suspend a process to perform a transformation or report statistics. The event marker feature is supported for the replication of data changes, but not for initial loads.

To trigger actions that do not require data to be applied to target tables, you can use the Replicat TABLE parameter with filtering options that support EVENTACTIONS. See "TABLE for Replicat" for more information.

Caution:

EVENTACTIONS is not supported if the source database is Teradata and Extract is configured in maximum performance mode.

You may need to combine two or more actions to achieve your goals. When multiple actions are combined, the entire EVENTACTIONS statement is parsed first, and then the specified options execute in order of precedence. The following list shows the order of precedence. The actions listed before Process the record occur before the record is written to the trail or applied to the target (depending on the process). Actions listed after Process the record are executed after the record is processed.

  • TRACE
  • LOG
  • CHECKPOINT BEFORE
  • DISCARD
  • SHELL
  • ROLLOVER
  • (Process the record)
  • IGNORE
  • REPORT
  • SUSPEND
  • ABORT
  • CHECKPOINT AFTER
  • FORCESTOP
  • STOP

To prevent the event record itself from being processed in the normal manner, use the IGNORE or DISCARD option. Because IGNORE and DISCARD are evaluated before the record itself, they prevent the record from being processed. Without those options, EVENTACTIONS for Extract writes the record to the trail, and EVENTACTIONS for Replicat applies that operation to the target database.

You should take into account the possibility that a transaction could contain two or more records that trigger an event action. In such a case, there could be multiple executions of certain EVENTACTIONS specifications. For example, encountering two qualifying records that trigger two successive ROLLOVER actions will cause Extract to roll over the trail twice, leaving one of the two files empty of transaction data.

You should also take into account that when the GETUPDATEBEFORES parameter is in effect, two records are generated for UPDATE operations: a record that contains the before image and a record that contains the after image. An event action is triggered for each of those records when the operation qualifies as an event record. You can use the BEFOREAFTERINDICATOR token of the GGHEADER column-conversion function as a filter in a FILTER clause to qualify the records so that the event action triggers only once, either on the before record or the after record, but not both.

The following example filters on the BEFORE indicator. The EVENTACTION issues the ECHO shell command to output the string 'Triggered on BEFORE' to the event log when a BEFORE record is encountered.

TABLE qasource.test, & 
FILTER(@STRFIND('BEFORE', @GETENV('GGHEADER' , 'BEFOREAFTERINDICATOR')) > 0), &
EVENTACTIONS ( shell ('echo --== Triggered on BEFORE ==-- '), LOG);
 

The following shows the result of the event action:

013-03-06 17:59:31  INFO    OGG-05301  Shell command output: '--== Triggered 
on AFTER ==--'
 

The following example does the same thing, but for the AFTER indicator.

TABLE qasource.test, & 
FILTER(@STRFIND('AFTER', @GETENV('GGHEADER' , 'BEFOREAFTERINDICATOR')) > 0), & 
EVENTACTIONS ( shell ('echo --== Triggered on AFTER ==-- '), LOG);

In a Teradata configuration where Extract is configured in maximum protection mode, use EVENTACTIONS only in the VAM-sort Extract group. It is not supported by the primary Extract in this configuration because concurrent changes are not sorted in transaction order at this point in the processing stream. For more information, see Installing and Configuring Oracle GoldenGate for Teradata.

Getting More Information About Configuring the Event Marker System

See Administering Oracle GoldenGate for Windows and UNIX for more information about using EVENTACTIONS and the Event Marker System.

Syntax

EVENTACTIONS (
[STOP | SUSPEND | ABORT | FORCESTOP]
[IGNORE [RECORD | TRANSACTION [INCLUDEVENT]]
[DISCARD]
[LOG [INFO | WARNING]]
[REPORT]
[ROLLOVER]
[SHELL 'command' |
  SHELL ('command', VAR variable = {column_name | expression}
  [, ...]) ]
[TRACE[2] file [TRANSACTION] [DDL[INCLUDE] | DDLONLY] [PURGE | APPEND]]
[CHECKPOINT [BEFORE | AFTER | BOTH]]
[, ...]
)
STOP

Valid in TABLE for Extract and in MAP for Replicat.

Brings the process to a graceful stop when the specified event record is encountered. The process waits for other operations within event transaction to be completed before stopping. If the transaction is a Replicat grouped or batched transaction, the current group of transactions are applied before the process stops gracefully. The process restarts at the next record after the event record, so long as that record also signified the end of a transaction.

The process logs a message if it cannot stop immediately because a transaction is still open. However, if the event record is encountered within a long-running open transaction, there is no warning message that alerts you to the uncommitted state of the transaction. Therefore, the process may remain running for a long time despite the STOP event.

STOP can be combined with other EVENTACTIONS options except for ABORT and FORCESTOP.

SUSPEND

Valid in TABLE for Extract and in MAP for Replicat.

Pauses the process so that it retains the active context of the current run and can still respond to SEND commands that are issued in GGSCI. When a process is suspended, the INFO command shows it as RUNNING, and the RBA field shows the last checkpoint position.

To resume processing, issue the SEND command with the RESUME option.

To use the CHECKPOINT BEFORE option in conjunction with SUSPEND, the event record must be the start of a transaction for the SUSPEND to take place. That way, if the process is killed while in the suspended state, the event record with the SUSPEND action is the first record to be reprocessed upon restart. If both CHECKPOINT BERORE and SUSPEND are specified, but the event record is not the start of a transaction, the process abends before SUSPEND can take place.

To use the CHECKPOINT AFTER option in conjunction with SUSPEND, the RESUME command must be issued before the checkpoint can take place, and the event record must be a COMMIT record. If the process is killed while in a SUSPEND state, the process reprocesses the transaction from the last checkpointed position upon restart.

SUSPEND cannot be combined with ABORT but can be combined with all other options.

ABORT

Valid in TABLE for Extract and in MAP for Replicat.

Forces the process to exit immediately when the specified event record is encountered, whether or not there are open transactions. The event record is not processed. A fatal error is written to the log, and the event record is written to the discard file if DISCARD is also specified. The process will undergo recovery on startup.

ABORT can be combined only with CHECKPOINT BEFORE, DISCARD, SHELL, and REPORT.

FORCESTOP

Valid in TABLE for Extract and in MAP for Replicat.

Forces the process to stop gracefully when the specified event record is encountered, but only if the event record is the last operation in the transaction or the only record in the transaction. The record is written normally.

If the event record is encountered within a long-running open transaction, the process writes a warning message to the log and exits immediately, as in ABORT. In this case, recovery may be required on startup. If the FORCESTOP action is triggered in the middle of a long-running transaction, the process exits without a warning message.

FORCESTOP can be combined with other EVENTACTIONS options except for ABORT, STOP, CHECKPOINT AFTER, and CHECKPOINT BOTH. If used with ROLLOVER, the rollover only occurs if the process stops gracefully.

IGNORE [RECORD | TRANSACTION [INCLUDEVENT]]

Valid in TABLE for Extract and in MAP for Replicat.

Ignores some or all of the transaction, depending on the selected action.

  • RECORD is the default. It forces the process to ignore only the specified event record, but not the rest of the transaction. No warning or message is written to the log, but the Oracle GoldenGate statistics are updated to show that the record was ignored.

  • Use TRANSACTION to ignore the entire transaction that contains the record that triggered the event. If TRANSACTION is used, the event record must be the first one in the transaction. When ignoring a transaction, the event record is also ignored by default. TRANSACTION can be shortened to TRANS.

  • Use INCLUDEEVENT with TRANSACTION to propagate the event record to the trail or to the target, but ignore the rest of the associated transaction.

IGNORE can be combined with all other EVENTACTIONS options except ABORT and DISCARD.

An IGNORE action is processed after all the qualification, filtering, mapping, and user-exit operations are processed. The record or transaction is ignored in the final output phase and prevents the record or transaction from being written to the output target (the trail in the case of Extract or the database in the case of Replicat). Therefore, in certain expressions, for example those that include SQLEXEC operations, the SQLEXEC will be executed before the IGNORE is processed. This means that, while the record is not written to the trail or target database, all of the effects of processing the record through qualification, filtering, mapping and user-exit will occur.

This action is not valid for DDL records. Because DDL operations are autonomous, ignoring a record is equivalent to ignoring the entire transaction.

DISCARD

Valid in TABLE for Extract and in MAP for Replicat.

Causes the process to:

  • write the specified event record to the discard file.

  • update the Oracle GoldenGate statistics to show that the record was discarded.

The process resumes processing with the next record in the trail.

DISCARD can be combined with all other EVENTACTIONS options except IGNORE.

LOG [INFO | WARNING]

Valid in TABLE for Extract and in MAP for Replicat.

Causes the process to log the event when the specified event record is encountered. The message is written to the report file, to the Oracle GoldenGate error log, and to the system event log.

Use the following options to specify the severity of the message:

  • INFO specifies a low-severity informational message. This is the default.

  • WARNING specifies a high-severity warning message.

LOG can be combined with all other EVENTACTIONS options except ABORT. If using ABORT, LOG is not needed because ABORT logs a fatal error before the process exits.

REPORT

Valid in TABLE for Extract and in MAP for Replicat.

Causes the process to generate a report file when the specified event record is encountered. This is the same as using the SEND command with the REPORT option in GGSCI.

The REPORT message occurs after the event record is processed (unless DISCARD, IGNORE, or ABORT are used), so the report data will include the event record.

REPORT can be combined with all other EVENTACTIONS options.

ROLLOVER

Valid in TABLE for Extract.

Causes Extract to roll over the trail to a new file when the specified event record is encountered. The ROLLOVER action occurs before Extract writes the event record to the trail file, which causes the record to be the first one in the new file unless DISCARD, IGNORE or ABORT are also used.

ROLLOVER can be combined with all other EVENTACTIONS options except ABORT. ROLLOVER cannot be combined with ABORT because ROLLOVER does not cause the process to write a checkpoint, and ROLLOVER happens before ABORT.

Without a ROLLOVER checkpoint, ABORT causes Extract to go to its previous checkpoint upon restart, which would be in the previous trail file. In effect, this cancels the rollover.

SHELL 'command'

Valid in TABLE for Extract and in MAP for Replicat.

Causes the process to execute the specified shell command when the event record is encountered. SHELL 'command' executes a basic shell command. The command string is taken at its literal value and sent to the system that way. The command is case-sensitive. Enclose the command string within single quote marks, for example:

EVENTACTIONS (SHELL 'echo hello world! > output.txt')

If the shell command is successful, the process writes an informational message to the report file and to the event log. Success is based upon the exit status of the command in accordance with the UNIX shell language. In that language, zero indicates success.

If the system call is not successful, the process abends with a fatal error. In the UNIX shell language, non-zero equals failure. Note that the error message relates only to the execution of the SHELL command itself, and not the exit status of any subordinate commands. For example, SHELL can execute a script successfully, but commands in that script could fail.

SHELL can be combined with all other EVENTACTIONS options.

SHELL ('command', VAR variable = {column_name | expression} [, ...])

Valid in TABLE for Extract and in MAP for Replicat.

Causes the process to execute the specified shell command when the event record is encountered and supports parameter passing. The command and the parameters are case-sensitive.

When SHELL is used with arguments, the entire command and argument strings must be enclosed within parentheses, for example:

EVENTACTIONS (SHELL ('Current timestamp: $1  SQLEXEC result is $2 ',VAR $1 = @GETENV('JULIANTIMESTAMP'),VAR $2 = mytest.description));

The input is as follows:

command

Is the command, which is passed literally to the system.

VAR

Is a required keyword that starts the parameter input.

variable

Is the user-defined name of the placeholder variable where the run-time variable value will be substituted. Extra variables that are not used in the command are ignored. Note that any literal in the SHELL command that matches a VAR variable name is replaced by the substituted VAR value. This may have unintended consequences, so test your code before putting it into production.

column_name

Can be the before or after (current) image of a column value.

expression

can be the following, depending on whether column data or DDL is being handled.

  • Valid expressions for column data:

    • The value from a TOKENS clause in a TABLE statement.

    • A return value from any Oracle GoldenGate column-conversion function.

    • A return value from a SQLEXEC query or procedure.

  • Valid expressions for DDL:

    • Return value from @TOKEN function (Replicat only).

    • Return value from @GETENV function.

    • Return value from other functions that do not reference column data (for example, @DATENOW).

    • Return value from @DDL function.

TRACE[2] file [TRANSACTION] [DDL[INCLUDE] | DDLONLY] [PURGE | APPEND]

Valid in TABLE for Extract and in MAP for Replicat.

Causes process trace information to be written to a trace file when the specified event record is encountered. TRACE provides step-by-step processing information. TRACE2 identifies the code segments on which the process is spending the most time.

By default (without options), standard DML tracing without consideration of transaction boundaries is enabled until the process terminates.

  • file specifies the name of the trace file and must appear immediately after the TRACE keyword. You can specify a unique trace file, or use the default trace file that is specified with the standalone TRACE or TRACE2 parameter.

    The same trace file can be used across different TABLE or MAP statements in which EVENTACTIONS TRACE is used. If multiple TABLE or MAP statements specify the same trace file name, but the TRACE options are not used consistently, preference is given to the options in the last resolved TABLE or MAP that contains this trace file.

  • Use TRANSACTION to enable tracing only until the end of the current transaction, instead of when the process terminates. For Replicat, transaction boundaries are based on the source transaction, not the typical Replicat grouped or batched target transaction. TRANSACTION can be shortened to TRANS. This option is valid only for DML operations.

  • DDL[INCLUDE] traces DDL and also DML transactional data processing. Either DDL or DDLINCLUDE is valid.

  • DDLONLY traces DDL but does not trace DML transactional data.

    These options are valid only for Replicat. By default DDL tracing is disabled.

  • Use PURGE to truncate the trace file before writing additional trace records, or use APPEND to write new trace records at the end of the existing records. APPEND is the default.

TRACE can be combined with all other EVENTACTIONS options except ABORT.

To disable tracing to the specified trace file, issue the GGSCI SEND process command with the TRACE OFF file_name option.

CHECKPOINT [BEFORE | AFTER | BOTH]

Valid in TABLE for Extract and in MAP for Replicat.

Causes the process to write a checkpoint when the specified event record is encountered. Checkpoint actions provide a context around the processing that is defined in TABLE or MAP statements. This context has a begin point and an end point, thus providing synchronization points for mapping the functions that are performed with SQLEXEC and user exits.

BEFORE

BEFORE for an Extract process writes a checkpoint before Extract writes the event record to the trail. BEFORE for a Replicat process writes a checkpoint before Replicat applies the SQL operation that is contained in the record to the target.

BEFORE requires the event record to be the first record in a transaction. If it is not the first record, the process will abend. Use BEFORE to ensure that all transactions prior to the one that begins with the event record are committed.

When using EVENTACTIONS for a DDL record, note that since each DDL record is autonomous, the DDL record is guaranteed to be the start of a transaction; therefore the CHECKPOINT BEFORE event action is implied for a DDL record.

CHECKPOINT BEFORE can be combined with all EVENTACTIONS options.

AFTER

AFTER for Extract writes a checkpoint after Extract writes the event record to the trail. AFTER for Replicat writes a checkpoint after Replicat applies the SQL operation that is contained in the record to the target.

AFTER flags the checkpoint request as an advisory, meaning that the process will only issue a checkpoint at the next practical opportunity. For example, in the case where the event record is one of a multi-record transaction, the checkpoint will take place at the next transaction boundary, in keeping with the Oracle GoldenGate data-integrity model.

When using EVENTACTIONS for a DDL record, note that since each DDL record is autonomous, the DDL record is guaranteed to be the end (boundary) of a transaction; therefore the CHECKPOINT AFTER event action is implied for a DDL record.

CHECKPOINT AFTER can be combined with all EVENTACTIONS options except ABORT.

BOTH

BOTH combines BEFORE and AFTER. The Extract or Replicat process writes a checkpoint before and after it processes the event record.

CHECKPOINT BOTH can be combined with all EVENTACTIONS options except ABORT.

CHECKPOINT can be shortened to CP.

Example 1   

The following example shows how you can configure a process to ignore certain records. When Extract processes any trail record that has name = abc, it ignores the record.

TABLE fin.cust, &
WHERE (name = 'abc'), &
EVENTACTIONS (ignore);
Example 2   

Based on the compatibility and precedence rules of EVENTACTIONS options, DISCARD takes higher precedence than ABORT, so in this example the event record gets written to the discard file before the process abends.

MAP fin.cust, TARGET fin.cust2, &
WHERE (name = 'abc'), &
EVENTACTIONS (DISCARD, ABORT);
Example 3   

The following example executes a SHELL action. It gets the result of a SQLEXEC query and pairs it with the current timestamp.

TABLE src.tab &
SQLEXEC (id mytest, query 'select description from lookup &
where pop = :mycol2', params (mycol2 = col2) ), &
EVENTACTIONS (SHELL ('Current timestamp: $1  SQLEXEC result is $2 ', &
VAR $1 = @GETENV('JULIANTIMESTAMP'), VAR $2 = mytest.description));

The shell command that results from this example could be similar to the following:

'Current timestamp: 212156002704718000  SQLEXEC result is test passed'
Example 4   

The following example shows how invalid results can occur if a placeholder name conflicts with literal text in the command string. In this example, a placeholder named $1 is associated with a column value, and the SHELL command echoes a literal string that includes $1.

MAP src.tab1, TARGET targ.tab1 &
EVENTACTIONS (SHELL ('echo Extra charge for $1 is $1', VAR $1 = COL1));

This is the unintended result, assuming the column value is gift wrap:

'Extra charge for gift wrap is gift wrap'

Changing the placeholder variable to $col results in the correct output:

MAP src.tab1, TARGET targ.tab1 &
EVENTACTIONS (SHELL ('echo Extra charge for $col is $1', VAR $col = COL1));
'Extra charge for gift wrap is $1'

The following shows similar potential for unintended results:

MAP src.tab1, TARGET targ.tab1 &
EVENTACTIONS (SHELL ('Timestamp: $1  Price is $13 > out.txt ', &
VAR $1 = @GETENV('JULIANTIMESTAMP')));

The redirected output file might contain a string like this (notice the second timestamp contains an appended value of 3):

'Timestamp: 212156002704718000 Price is 2121560027047180003'

The intended result is this:

'Timestamp: 212156002704718000 Price is $13'
Example 5   

These examples show different ways to configure tracing.

MAP tab1, TARGET tab1 EVENTACTIONS (TRACE ./dirrpt/trace1.txt);
MAP tab2, TARGET tab2 EVENTACTIONS (TRACE ./dirrpt/trace2.txt TRANSACTION);
  • In the first MAP statement, the trace1.txt trace file is generated just before the first tab1 event record is applied to the target. It contains all of the tracing information from that point forward until Replicat terminates or unless tracing is turned off with the GGSCI SEND REPLICAT command.

  • Because the second MAP statement contains the TRANSACTION option, the trace2.txt file is generated just before the first tab2 event record is applied to the target, but the tracing stops automatically at the conclusion of the transaction that contains the tab2 event record.

Example 6   

The following shows how EVENTACTIONS with SUSPEND can be used.

  • Case 1: You are replicating DDL, and you want to ensure that there is enough space in the target database to create a new table. Use EVENTACTIONS with SUSPEND in the MAP statement that maps the CREATE TABLE DDL operation, and then execute a SQL statement in that MAP statement to query the amount of space remaining in a tablespace. If there is enough space, use SEND REPLICAT with RESUME to resume processing immediately; if not, leave Replicat suspended until a DBA can add the space, and then use SEND REPLICAT with RESUME to resume processing.

  • Case 2: You want to fix unique key violations when they occur on any table. Because Replicat is processing thousands of tables, you do not want to stop the process each time there is a violation, because this would cause Replicat to spend time rebuilding the object cache again upon restart. By using EVENTACTIONS with SUSPEND, you can simply suspend processing until the problem is fixed.

  • Case 3: At the end of the day, you suspend Replicat to run daily reports, and then resume processing immediately without stopping and restarting the process.

EXITPARAM 'parameter'

EXITPARAM is valid for TABLE and MAP.

Use EXITPARAM to pass a parameter to the EXIT_PARAMS function of a user exit routine whenever a record from the TABLE or MAP statement is encountered.

Getting More Information about User Exits

See Administering Oracle GoldenGate for Windows and UNIX for instructions on how to configure user exits.

See "User Exit Functions" for more information about the syntax for the user exits.

Syntax

EXITPARAM 'parameter string'
'parameter string'

A parameter that is a literal string. Enclose the parameter within single quotes. You can specify up to 100 characters for the parameter string.

FETCHBEFOREFILTER

FETCHBEFOREFILTER is valid for TABLE.

Use FETCHBEFOREFILTER to fetch columns that are specified with FETCHCOLS or FETCHCOLSEXCEPT before a FILTER operation is executed. Fetching before the filter ensures that values required for the filter are available. Without FETCHBEFOREFILTER, fetches specified with FETCHCOLS or FETCHCOLSEXCEPT are not performed until after filters are executed. Specify FETCHBEFOREFILTER before FILTER in the parameter file.

Do not use this option for tables being processed in pass-through mode by a data-pump Extract group.

FETCHBEFOREFILTER is not supported for the SQL/MX database.

Syntax

FETCHBEFOREFILTER

Example

TABLE hr.salary, FETCHCOLS (sal_level),
FETCHBEFOREFILTER,
FILTER (sal_level >= 8)
;

{FETCHCOLS | FETCHCOLSEXCEPT} (column_list)

FETCHCOLS and FETCHCOLSEXCEPT are valid for TABLE. These options are only valid for the primary extract and cannot be used on data pump.

Use FETCHCOLS and FETCHCOLSEXCEPT to fetch column values from the database when the values are not present in the transaction log record. Use this option if the database only logs the values of columns that were changed in an update operation, but you need to ensure that other column values required for FILTER operations are available.

  • FETCHCOLS fetches the specified columns.

  • FETCHCOLSEXCEPT fetches all columns except the specified columns. For tables with numerous columns, FETCHCOLSEXCEPT may be more efficient than listing each column with FETCHCOLS.

FETCHCOLS and FETCHCOLSEXCEPT are valid for all databases that are supported by Oracle GoldenGate, except NonStop SQL/MX.

For an Oracle database, Oracle GoldenGate fetches the values from the undo tablespace through Oracle's Flashback Query mechanism. The query provides a read-consistent image of the columns as of a specific time or SCN. For more information about how Oracle GoldenGate uses Flashback Query, see Installing and Configuring Oracle GoldenGate for Oracle Database.

Instead of using FETCHCOLS or FETCHCOLSEXCEPT, it may be more efficient to enable supplemental logging for the desired columns.

For Sybase, encrypted column data is not supported by these parameters because Oracle GoldenGate does not support Sybase encrypted data.

To control fetching and enable a response when a column specified for fetching cannot be located, use the FETCHOPTIONS parameter. To include fetch results in statistical displays generated by the STATS EXTRACT command, use the STATOPTIONS parameter.

If values for columns specified with FETCHCOLS or FETCHCOLSEXCEPT are present in the transaction log, no database fetch is performed. This reduces database overhead.

Syntax

{FETCHCOLS | FETCHCOLSEXCEPT} (column [, ...])
column

Can be one of the following:

  • A column name or a comma-delimited list of column names, as in (col1, col2).

  • An asterisk wildcard, as in (*).

Example

The FETCHCOLS clause in this example fetches only columns 1 and 3, whereas the FETCHCOLSEXCEPT clause fetches all columns except columns 1 and 3.

TABLE hq.acct, FETCHCOLS (col1, col3);
TABLE hq.sales, FETCHCOLSEXCEPT (col1, col3);

{FETCHMODCOLS | FETCHMODCOLSEXCEPT} (column_list)

FETCHMODCOLS and FETCHMODCOLSEXCEPT are valid for TABLE. These options are only valid for the primary extract and cannot be used on data pump.

Use FETCHMODCOLS and FETCHMODCOLSEXCEPT to force column values to be fetched from the database even if the columns are present in the transaction log. These Depending on the database type, a log record can contain all of the columns of a table or only the columns that changed in the given transaction operation.

  • FETCHMODCOLS fetches the specified columns.

  • FETCHMODCOLSEXCEPT fetches all columns that are present in the transaction log, except the specified columns. For tables with numerous columns, FETCHMODCOLSEXCEPT might be more efficient than listing each column with FETCHMODCOLS.

FETCHMODCOLS and FETCHMODCOLSEXCEPT are valid for all databases that are supported by Oracle GoldenGate, except NonStop SQL/MX.

Observe the following usage guidelines:

  • Do not use FETCHMODCOLS and FETCHMODCOLSEXCEPT for key columns.

  • (Sybase) Do not use FETCHMODCOLS and FETCHMODCOLSEXCEPT for encrypted column data. Oracle GoldenGate does not support Sybase encrypted data.

Syntax

{FETCHMODCOLS | FETCHMODCOLSEXCEPT} (column [, ...])
(column [, ...])

Can be one of the following:

  • A column name or a comma-delimited list of column names, as in (col1, col2).

  • An asterisk wildcard, as in (*).

Example

The FETCHMODCOLS clause in this example fetches only columns 1 and 3, whereas the FETCHMODCOLSEXCEPT clause fetches all columns except columns 1 and 3.

TABLE hq.acct, FETCHMODCOLS (col1, col3);
TABLE hq.sales, FETCHMODCOLSEXCEPT (col1, col3);

FILTER (filter_clause)

FILTER is valid for TABLE and MAP.

Use FILTER to select or exclude records based on a numeric value. A filter expression can use conditional operators, Oracle GoldenGate column-conversion functions, or both.

Note:

To filter based on a string, use one of the Oracle GoldenGate string functions. See "Column Conversion Functions" for more information about these functions. You can also use the WHERE option. See "WHERE (clause)".

Separate all FILTER components with commas. A FILTER clause can include the following:

  • Numbers

  • Columns that contain numbers

  • Functions that return numbers

  • Arithmetic operators:

    + (plus)

    - (minus)

    * (multiply)

    / (divide)

    \ (remainder)

  • Comparison operators:

    > (greater than)

    >= (greater than or equal)

    < (less than)

    <= (less than or equal)

    = (equal)

    <> (not equal)

    Results derived from comparisons can be zero (indicating FALSE) or non-zero (indicating TRUE).

  • Parentheses (for grouping results in the expression)

  • Conjunction operators: AND, OR

Enclose literals in single quotes. Specify case-sensitive column names as they are stored in the database, and enclose them in double quotes if the database requires quotes to enforce case-sensitivity (such as Oracle).

Oracle GoldenGate supports FILTER for columns that have a multi-byte character set.

Getting More Information about Record Filtering

See Administering Oracle GoldenGate for Windows and UNIX for more information about FILTER and other filtering options.

Syntax

FILTER (
[, ON INSERT | ON UPDATE| ON DELETE]
[, IGNORE INSERT | IGNORE UPDATE | IGNORE DELETE]
, filter_clause
[, RAISEERROR error_number]
)
filter_clause

Selects records based on an expression, such as:

FILTER ((PRODUCT_PRICE*PRODUCT_AMOUNT) > 10000))

You can use the column-conversion functions of Oracle GoldenGate in a filter clause, as in:

FILTER (@COMPUTE (PRODUCT_PRICE*PRODUCT_AMOUNT)>10000)

Enclose literals in single quotes. Specify case-sensitive column names as they are stored in the database, and enclose them in double quotes if the database requires quotes to enforce case-sensitivity (such as Oracle).

Oracle GoldenGate does not support FILTER for columns that have a multi-byte character set or a character set that is incompatible with the character set of the local operating system.

The maximum size of the filter clause is 5,000 bytes.

ON INSERT | ON UPDATE | ON DELETE

Restricts record filtering to the specified operation(s). Separate operations with commas, for example:

FILTER (ON UPDATE, ON DELETE,
@COMPUTE (PRODUCT_PRICE*PRODUCT_AMOUNT)>10000)

The preceding example executes the filter for UPDATE and DELETE operations, but not INSERT operations.

IGNORE INSERT | IGNORE UPDATE | IGNORE DELETE

Does not apply the filter for the specified operation(s). Separate operations with commas, for example:

FILTER (IGNORE INSERT, @COMPUTE (PRODUCT_PRICE*PRODUCT_AMOUNT)>10000)

The preceding example executes the filter on UPDATE and DELETE operations, but ignores INSERT operations.

RAISEERROR error

Raises a user-defined error number if the filter fails. Can be used as input to the REPERROR parameter to invoke error handling. Make certain that the value for error is outside the range of error numbers that is used by the database or by Oracle GoldenGate. For example: RAISEERROR 21000.

GETBEFORECOLS (column_specification)

GETBEFORECOLS is valid for TABLE.

Use GETBEFORECOLS to specify columns for which you want before image to be captured and written to the trail upon an update or delete operation. Use GETBEFORECOLS when using the Oracle GoldenGate Conflict Detection and Resolution (CDR) feature in a bi-directional or multi-master configuration. Also use it when using conversion functions or other processing features that require the before image of a record.

For updates, the before image of the specified columns is included in the trail whether or not any given column is modified. In addition to the columns specified in the GETBEFORECOLS clause, an Oracle database will also log the before image of other columns that are modified. For other supported databases, you can use the GETUPDATEBEFORES parameter to force the inclusion of the before values of other columns that are modified.

Note:

GETUPDATEBEFORES overrides GETBEFORECOLS if both are used in the same parameter file.

To use this parameter, supplemental logging must be enabled for any database that does not log before values by default.

GETBEFORECOLS overrides COMPRESSUPDATES and COMPRESSDELETES if used in the same parameter file.

This parameter is valid for all databases except DB2. For DB2 on all platforms that are supported by Oracle GoldenGate, use the GETUPDATEBEFORES parameter instead of GETBEFORECOLS.

Syntax

GETBEFORECOLS(
{ON UPDATE | ON DELETE}
{ALL | KEY | KEYINCLUDING (col[,...])  | KEYANDMOD | | ALLEXCLUDING (col[,...]) }
[,...]
)
{ON UPDATE | ON DELETE}

Specifies whether the before image of the specified columns should be captured for updates or deletes. You can use ON UPDATE only, ON DELETE only, or both. If using both, specify them within the same GETBEFORECOLS clause. See the example for how to use both.

{ALL | KEY | KEYINCLUDING (col[,...]) | KEYANDMOD | ALLEXCLUDING (col[,...])}

Specifies the columns for which a before image is captured.

ALL

Captures a before image of all supported data type columns in the target table, including the primary key; all unsupported columns are skipped and logged in the Extract or Replicat parameter file as an information message. This imposes the highest processing load for Extract, but allows conflict-detection comparisons to be performed using all columns for maximum accuracy.

KEY

Capture before image only for the primary key. This is the fastest option, but does not permit the most accurate conflict detection, because keys can match but non-key columns could be different. KEY is the default.

KEYINCLUDING

Capture before image of the primary key and also the specified column or columns. This is a reasonable compromise between speed and detection accuracy.

KEYANDMOD

Use this option as an extension of the key option for both Extract and Replicat. For update DMLs on the source, Extract logs the key and modified columns. Replicat on the target will use the KEY and MODIFIED columns during conflict detection in a WHERE clause. With Oracle databases, the modified column is always used for conflict detection by default and this parameter makes it explicit.

ALLEXCLUDING

Capture before image of all columns except the specified columns. For tables with numerous columns, ALLEXCLUDING may be more efficient than KEYINCLUDING. Do not exclude key columns.

Example

In the following example, the before images for the key column(s) plus the name, address, and salary are always written to the trail file on update and delete operations.

TABLE src,
GETBEFORECOLS (
ON UPDATE KEYINCLUDING (name, address,  salary),
ON DELETE KEYINCLUDING (name, address, salary));

HANDLECOLLISIONS | NOHANDLECOLLISIONS

HANDLECOLLISIONS and NOHANDLECOLLISIONS are valid for MAP.

Use HANDLECOLLISIONS and NOHANDLECOLLISIONS to control whether or not Oracle GoldenGate reconciles the results of an initial load with replicated transactional changes that are made to the same tables. When Oracle GoldenGate applies replicated changes after the load is finished, HANDLECOLLISIONS causes Replicat to overwrite duplicate records in the target tables and provides alternate handling of errors for missing records.

HANDLECOLLISIONS and NOHANDLECOLLISIONS can be used globally for all MAP statements in the parameter file or as an ON/OFF switch for groups of tables specified with MAP statements, and they can be used within a MAP statement. When used in a MAP statement, they override the global specifications.

See "HANDLECOLLISIONS | NOHANDLECOLLISIONS" for syntax and usage.

INSERTALLRECORDS

INSERTALLRECORDS is valid for MAP.

Use the INSERTALLRECORDS parameter to convert all mapped operations to INSERT operations on the target. INSERTALLRECORDS can be used at the root level of the parameter file, within a MAP statement, and within a MAPEXCEPTION clause of a MAP statement.

See "INSERTALLRECORDS" for syntax and usage.

INSERTAPPEND | NOINSERTAPPEND

INSERTAPPEND is valid for MAP.

Use the INSERTAPPEND and NOINSERTAPPEND parameters to control whether or not Replicat uses an APPEND hint when it applies INSERT operations to Oracle target tables. These parameters are valid only for Oracle databases.

See "INSERTAPPEND | NOINSERTAPPEND" for syntax and usage.

KEYCOLS (columns)

KEYCOLS is valid for TABLE and MAP.

Use KEYCOLS to define one or more columns of the target table as unique. The primary use for KEYCOLS is to define a substitute primary key when a primary key or an appropriate unique index is not available for the table. You can also use KEYCOLS to specify additional columns to use in the row identifier that Replicat uses. Without the availability of a key or KEYCOLS clause, Replicat uses all columns of the table to build its WHERE clause, essentially performing a full table scan.

The columns of a key rendered by KEYCOLS must uniquely identify a row, and they must match the columns that are used as a key on the source table. The source table must contain at least as many key or index columns as the KEYCOLS key specified for the target table. Otherwise, in the event of an update to the source key or index columns, Replicat will not have the before images for the extra target KEYCOL columns.

When defining a substitute key with KEYCOLS, observe the following guidelines:

  • If the source and target tables both lack keys or unique indexes, use a KEYCOLS clause in the TABLE parameter and in the MAP parameter, and specify matching sets of columns in each KEYCOLS clause.

  • If either of the tables lacks a key or unique index, use KEYCOLS for that table. Specify columns that match the actual key or index columns of the other table. If a matching set cannot be defined with KEYCOLS, you must use KEYCOLS for the source table (TABLE parameter) and for the target table (MAP parameter). Specify matching sets of columns that contain unique values. KEYCOLS overrides a key or unique index.

  • If the target table has a larger key than the source table does (or if it has more unique-index columns), use KEYCOLS in the TABLE statement to specify the source columns that match the extra target columns. You must also include the actual source key or index columns in this KEYCOLS clause. Using KEYCOLS in this way ensures that before images are available to Replicat in case the non-key columns are updated on the source.

When using KEYCOLS, make certain that the specified columns are configured for logging so that they are available to Replicat in the trail records. For an Oracle database, you can enable the logging by using the COLS option of the ADD TRANDATA command.

On the target tables, create a unique index on the KEYCOLS-defined key columns. An index improves the speed with which Oracle GoldenGate locates the target rows that it needs to process.

Do not use KEYCOLS for tables being processed in pass-through mode by a data-pump Extract group.

Syntax

KEYCOLS (column [, ... ])
column

Defines a column to be used as a substitute primary key. If a primary or unique key exists, those columns must be included in the KEYCOLS specification. To specify multiple columns, create a comma-delimited list as in:

KEYCOLS (id, name)

The following column-types are not supported in KEYCOLS:

  • Oracle column types not supported by KEYCOLS:

    Virtual columns, UDTs, function-based columns, and any columns that are explicitly excluded from the Oracle GoldenGate configuration.

  • SQL Server, DB2 LUW, DB2 z/OS, MySQL, SQL/MX, Teradata, TimesTen column types not supported by KEYCOLS:

    Columns that contain a timestamp or non-materialized computed column, and any columns excluded from the Oracle GoldenGate configuration. For SQL Server Oracle GoldenGate enforces the total length of data in rows for target tables without a primary key to be below 8000 bytes.

  • Sybase column types not supported by KEYCOLS:

    Computed columns, function-based columns, and any columns that are explicitly excluded from the GoldenGate configuration.

Example

TABLE hr.emp, KEYCOLS (id, first, last, birthdate);

MAPEXCEPTION (exceptions_mapping)

MAPEXCEPTIONS is valid for MAP.

Use MAPEXCEPTION as part of an exceptions MAP statement intended for error handling. MAPEXCEPTION maps failed operations that are flagged as exceptions by the REPERROR parameter to an exceptions table. Replicat writes the values of these operations along with other information to the exceptions table.

You can use MAPEXCEPTION within the same MAP statement that includes the source-target table mapping and other standard MAP options. The source and target table names can include wildcards.

When using MAPEXCEPTION, use a REPERROR statement with the EXCEPTION option either within the same MAP statement or at the root of the Replicat parameter file. See "EXCEPTIONSONLY" and "REPERROR".

Getting More Information About Exceptions Handling

For more information about configuring exceptions handling with an exceptions MAP statement, see Administering Oracle GoldenGate for Windows and UNIX.

Syntax

MAPEXCEPTION (TARGET exceptions_table, INSERTALLRECORDS [, exception_MAP_options])
TARGET exceptions_table

The fully qualified name of the exceptions table. Standard Oracle GoldenGate rules for object names apply to the name of the exceptions table. See Administering Oracle GoldenGate for Windows and UNIX.

exception_MAP_options

Any valid options of the MAP parameter that you want to apply to the exceptions handling.

INSERTALLRECORDS

Applies all exceptions to the exceptions table as INSERT operations. This parameter is required when using MAPEXCEPTION.

Example

This is an example of how to use MAPEXCEPTION for exceptions mapping. The MAP and TARGET clauses contain wildcarded source and target table names. Exceptions that occur when processing any table with a name beginning with TRX will be captured to the fin.trxexceptions table using the specified mapping.

MAP src.trx*, TARGET trg.*,
MAPEXCEPTION (TARGET fin.trxexceptions,
INSERTALLRECORDS,
COLMAP (USEDEFAULTS,
ACCT_NO = ACCT_NO,
OPTYPE = @GETENV ('LASTERR', 'OPTYPE'),
DBERR = @GETENV ('LASTERR', 'DBERRNUM'),
DBERRMSG = @GETENV ('LASTERR', 'DBERRMSG')
)
);

MAPINVISIBLECOLUMNS | NOMAPINVISIBLECOLUMNS

MAPINVISIBLECOLUMNS and NOMAPINVISIBLECOLUMNS are valid for MAP.

Use MAPINVISIBLECOLUMNS and NOMAPINVISIBLECOLUMNS to control whether or not Replicat includes invisible columns in Oracle target tables for default column mapping. For invisible columns in Oracle target tables that use explicit column mapping, they are always mapped so do not require this option.

MAPINVISIBLECOLUMNS and NOMAPINVISIBLECOLUMNS can be used in two different ways. When specified at a global level, one parameter remains in effect for all subsequent MAP statements, until the other parameter is specified. When used within a MAP statement, they override the global specifications

See “MAPINVISIBLECOLUMNS | NOMAPINVISIBLECOLUMNS” for syntax and usage.

REPERROR (error, response)

REPERROR is valid for MAP.

Use REPERROR to specify an error and a response that together control how Replicat responds to the error when executing the MAP statement. You can use REPERROR at the MAP level to override and supplement global error handling rules set with the REPERROR parameter at the root level of the parameter file. Multiple REPERROR statements can be applied to the same MAP statement to enable automatic, comprehensive management of errors and interruption-free replication processing.

For syntax and descriptions, see "REPERROR".

RESOLVECONFLICT (conflict_resolution_specification)

RESOLVECONFLICT is valid for MAP.

Use RESOLVECONFLICT in a bi-directional or multi-master configuration to specify how Replicat handles conflicts on operations made to the tables in the MAP statement.

Multiple resolutions can be specified for the same conflict type and are executed in the order listed in RESOLVECONFLICT. Multiple resolutions are limited to INSERTROWEXISTS and UPDATEROWEXISTS conflicts only.

RESOLVECONFLICT can be used multiple times in a MAP statement to specify different resolutions for different conflict types.

The following are the data types and platforms that are supported by RESOLVECONFLICT.

  • RESOLVECONFLICT supports all databases that are supported by Oracle GoldenGate for Windows and UNIX.

  • To use RESOLVECONFLICT, the database must reside on a Windows, Linux, or UNIX system (including those running on NonStop OSS).

  • CDR supports data types that can be compared with simple SQL and without explicit conversion. See the individual parameter options for details.

  • Do not use RESOLVECONFLICT for columns that contain LOBs, abstract data types (ADT), or user-defined types (UDT).

  • Do not use RESOLVECONFLICT for BigNum data types in a SQL/MX database.

Getting More Information About Configuring Conflict Resolution

See Administering Oracle GoldenGate for Windows and UNIX for detailed instructions on configuring bi-directional replication and conflict resolution, including use cases and examples.

Syntax

RESOLVECONFLICT (
{INSERTROWEXISTS | UPDATEROWEXISTS | UPDATEROWMISSING | 
   DELETEROWEXISTS | DELETEROWMISSING} 
( {DEFAULT | resolution_name},
  {USEMAX (resolution_column) | USEMAXEQ (resolution_column) | USEMIN (resolution_column) | USEMINEQ (resolution_column) | USEDELTA |
     DISCARD | OVERWRITE | IGNORE} 
 )   
[, COLS (column[,...])]
)
INSERTROWEXISTS | UPDATEROWEXISTS | UPDATEROWMISSING |
DELETEROWEXISTS | DELETEROWMISSING

The type of conflict that this resolution handles.

INSERTROWEXISTS

An inserted row violates a uniqueness constraint on the target.

UPDATEROWEXISTS

An updated row exists on the target, but one or more columns have a before image in the trail that is different from the current value in the database.

UPDATEROWMISSING

An updated row does not exist in the target.

DELETEROWEXISTS

A deleted row exists in the target, but one or more columns have a before image in the trail that is different from the current value in the database.

DELETEROWMISSING

A deleted row does not exist in the target.

DEFAULT | resolution_name
DEFAULT

The default column group. The resolution that is associated with the DEFAULT column group is used for all columns that are not in an explicitly named column group. You must define a DEFAULT column group.

resolution_name

A name for a specific column group that is linked to a specific resolution type. Supply a name that identifies the resolution type. Valid values are alphanumeric characters. Avoid spaces and special characters, but underscores are permitted, for example:

delta_res_method

Use either a named resolution or DEFAULT, but not both.

USEMAX (resolution_column) | USEMAXEQ (resolution_column) | USEMIN (resolution_column) | USEMINEQ (resolution_column) | USEDELTA |
DISCARD | OVERWRITE | IGNORE

The conflict-handler logic that is used to resolve the conflict. Valid resolutions are:

USEMAX

If the value of resolution_column in the trail record is greater than the value of the column in the database, the appropriate action is performed.

  • (INSERTROWEXISTS conflict) Apply the trail record, but change the insert to an update to avoid a uniqueness violation, and overwrite the existing values.

  • (UPDATEROWEXISTS conflict) Apply the trail record as an update.

USEMAXEQ

If the value of resolution_column in the trail record is greater than or equal to the value of the column in the database, the appropriate action is performed.

  • (INSERTROWEXISTS conflict) Apply the trail record, but change the insert to an update to avoid a uniqueness violation, and overwrite the existing values.

  • (UPDATEROWEXISTS conflict) Apply the trail record as an update.

USEMIN

If the value of resolution_column in the trail record is less than the value of the column in the database, the appropriate action is performed:

  • (INSERTROWEXISTS conflict) Apply the trail record, but change the insert to an update to avoid a uniqueness violation, and overwrite the existing values.

  • (UPDATEROWEXISTS conflict) Apply the update from the trail record.

USEMINEQ

If the value of resolution_column in the trail record is less than or equal to the value of the column in the database, the appropriate action is performed:

  • (INSERTROWEXISTS conflict) Apply the trail record, but change the insert to an update to avoid a uniqueness violation, and overwrite the existing values.

  • (UPDATEROWEXISTS conflict) Apply the update from the trail record.

resolution_column

The name of a NOT NULL column that serves as the resolution column. This column must be part of the column group that is associated with this resolution. The value of the resolution column compared to the current value in the target database determines how a resolution should be applied. The after image of the resolution column is used for the comparison, if available; otherwise the before image value is used. Use a column that can be compared through simple SQL:

  • NUMERIC

  • DATE

  • TIMESTAMP

  • CHAR/NCHAR

  • VARCHAR/ NVARCHAR

To use a latest-timestamp resolution, use a timestamp column as the resolution_column and set the timestamp column to the current time when a row is inserted or updated. If possible, define the resolution column with the SYSTIMESTAMP data type, which supports fractional seconds. When comparisons are performed with sub-second granularity, there is little need for tie-breaking conflict handlers that resolve cases where the value of the resolution column is identical in both trail and target. If you ensure that the value of the timestamp column can only increase or only decrease (depending on the resolution), then USEMAX and USEMIN does not lead to data divergence.

Note:

Do not use a primary key column as the resolution column in a USEMAX statement for the UPDATEROWEXISTS conflict. Otherwise, Replicat abends with an error similar to the following:

2013-04-04 10:18:38  ERROR   OGG-01922  Missing  RESOLUTION COLUMN NAME while mapping to target table "FIN"."ACCT".
USEDELTA

(UPDATEROWEXISTS conflict only) Add the difference between the before and after values in the trail record to the current value of the column in the target database. If any of the values is NULL, an error is raised. Base USEDELTA on columns that contain NUMERIC data types. USEDELTA is useful in a multi-node configuration when a row is getting simultaneously updated on multiple nodes. It propagates only the difference in the column values to the other nodes, so that all nodes become synchronized.

DISCARD

(Valid for all conflict types) Retain the current value in the target database, and write the data in the trail record to the discard file.

Use DISCARD with caution, because it can lead to data divergence.

OVERWRITE

(Valid for all conflict types except DELETEROWMISSING) Apply the trail record as follows:

  • (INSERTROWEXISTS conflict) Apply the trail record but change the insert to an update to avoid a uniqueness violation, and overwrite the existing values.

  • (UPDATEROWEXISTS conflict) Apply the update from the trail record.

  • (UPDATEROWMISSING conflict) Apply the trail record but convert the missing UPDATE to an INSERT by using the modified columns from the after image and the unmodified columns from the before image. To convert an update to an insert, the before image of all columns of the row must be available in the trail. Use supplemental logging if the database does not log before images by default, and specify ALL for the Extract GETBEFORECOLS parameter.

  • (DELETEROWEXISTS conflict) Apply the delete from the trail record, but use only the primary key columns in the WHERE clause.

Use OVERWRITE with caution, because it can lead to data divergence.

IGNORE

(Valid for all conflict types) Retain the current value in the target database, and ignore the trail record: Do not apply to the target table or a discard file.

COLS (column[, ...])

A non-default column group. This is a list of columns in the target database (after mapping) that are linked to, and operated upon by, a specific resolution type. If no column group is specified for a conflict, then all columns are affected by the resolution that is specified for the given conflict.

Alternatively, you can specify a DEFAULT column group, which includes all columns that are not listed in another column group. See the DEFAULT option.

You can specify multiple column groups, each with a different resolution. For example, you could use OVERWRITE for col2 and col3, and you could use USEDELTA for col4. No column in any group can be in any other group. Conflicts for columns in different column groups are resolved separately according to the specified resolution, and in the order listed.

Column groups work as follows:

  • For INSERTROWEXISTS and UPDATEROWEXISTS conflicts, you can use different column groups to specify more than one of these conflict types and resolutions per table. Conflicts for columns in different column groups are resolved separately, according to the conflict resolution method specified for the column group.

  • For UPDATEROWMISSING, DELETEROWEXISTS, and DELETEROWMISSING, you can use only one column group, and all columns of the table must be in this column group (considered the default column group).

Examples

The following examples are explained in detail in Administering Oracle GoldenGate for Windows and UNIX.

Example 1   

This example demonstrates all conflict types with USEMAX, OVERWRITE, DISCARD.

MAP fin.src, TARGET fin.tgt,
    COMPARECOLS (ON UPDATE ALL, ON DELETE ALL),
    RESOLVECONFLICT (UPDATEROWEXISTS, (DEFAULT, USEMAX (last_mod_time)),
    RESOLVECONFLICT (INSERTROWEXISTS, (DEFAULT, USEMAX (last_mod_time)),
    RESOLVECONFLICT (DELETEROWEXISTS, (DEFAULT, OVERWRITE)),
    RESOLVECONFLICT (UPDATEROWMISSING, (DEFAULT, OVERWRITE)),
    RESOLVECONFLICT (DELETEROWMISSING, (DEFAULT, DISCARD)),
    );
Example 2   

This example demonstrates UPDATEROWEXISTS with USEDELTA and USEMAX.

MAP fin.src, TARGET fin.tgt,
    COMPARECOLS
    (ON UPDATE KEYINCLUDING (address, phone, salary, last_mod_time),
    ON DELETE KEYINCLUDING (address, phone, salary, last_mod_time)),
    RESOLVECONFLICT (
    UPDATEROWEXISTS,
    (delta_res_method, USEDELTA, COLS (salary)),
    (DEFAULT, USEMAX (last_mod_time)));
Example 3   

This example demonstrates UPDATEROWEXISTS with USEDELTA, USEMAX, and IGNORE.

MAP fin.src, TARGET fin.tgt,
    COMPARECOLS
    (ON UPDATE ALLEXCLUDING (comment)),
    RESOLVECONFLICT (
    UPDATEROWEXISTS,
    (delta_res_method, USEDELTA, COLS (salary, balance)),
    (max_res_method, USEMAX (last_mod_time), COLS (address, last_mod_time)),
    (DEFAULT, IGNORE));

SQLEXEC (SQL_specification)

SQLEXEC is valid for TABLE and MAP.

Use SQLEXEC to execute a SQL stored procedure or query from within a MAP statement during Oracle GoldenGate processing. SQLEXEC enables Oracle GoldenGate to communicate directly with the database to perform any work that is supported by the database. This work can be part of the synchronization process, such as retrieving values for column conversion, or it can be independent of extracting or replicating data, such as executing a stored procedure that executes an action within the database.

See "SQLEXEC" for syntax and usage.

SQLPREDICATE 'WHERE where_clause'

SQLPREDICATE is valid for TABLE.

Use SQLPREDICATE to include a conventional SQL WHERE clause in the SELECT statement that Extract uses when selecting data from a table in preparation for an initial load. SQLPREDICATE forces the records returned by the selection to be ordered by the key values.

SQLPREDICATE is a faster selection method for initial loads than the WHERE or FILTER options. It affects the SQL statement directly and does not require Extract to fetch all records before filtering them.

For Oracle tables, SQLPREDICATE reduces the amount of data that is stored in the undo segment, which can reduce the incidence of snapshot-too-old errors. This is useful when loading very large tables.

By using a SQLPREDICATE clause, you can partition the rows of a large table among two or more parallel Extract processes. This configuration enables you to take advantage of parallel delivery load processing as well.

SQLPREDICATE also enables you to select data based on a timestamp or other criteria to filter the rows that are extracted and loaded to the target table. SQLPREDICATE can be used for ORDER BY clauses or any other type of selection clause.

Make certain that the WHERE clause contains columns that are part of a key or index. Otherwise, Extract performs a full table scan, which reduces the efficiency of the SELECT statement.

SQLPREDICATE is valid for Oracle, DB2 LUW, DB2 on z/OS, SQL Server, and Teradata databases. Do not use SQLPREDICATE for an Extract group that is configured to synchronize transactional changes. It is only appropriate for an initial load Extract, because it re quires a SELECT statement that selects records directly from tables.

Syntax

TABLE source_table, SQLPREDICATE 'WHERE where_clause';
WHERE

This is a required keyword.

where_clause

A valid SQL WHERE clause that selects records from the source tables.

Example

TABLE hr.emp, SQLPREDICATE 'WHERE state = 'CO' and city = 'DENVER''

THREAD (thread_ID)

THREAD is valid for MAP. This option is valid when Replicat is in coordinated mode.

Use THREAD to specify that all of the object or objects in the same MAP statement are to be processed by the specified Replicat thread. The specified thread handles filtering, manipulation, delivery to the target, error handling, and other work that is configured for those objects. Wildcards can be used in the TARGET clause when THREAD is used.

All tables that have referential dependencies among one another must be mapped in the same thread. For example, if tables scott.cust and scott.ord have a foreign-key relationship, the following is a possible mapping:

MAP scott.cust, TARGET scott.cust, THREAD (5);
MAP scott.ord, TARGET scott.ord, THREAD (5);

The thread with the lowest thread ID always processes barrier transactions if the THREAD or THREADRANGE option is omitted. Additionally, and work that is not explicitly assigned to a thread is processed through this thread. For example, if there are threads with IDs ranging from 1 to 10, barrier and non-assigned transactions are performed by thread 1.

To process a MAP statement among multiple threads, see THREADRANGE (thread_range, column_list). THREAD and THREADRANGE are mutually exclusive options. Do not use them together in the same MAP statement.

For more information about Replicat modes, see "Deciding Which Apply Method to Use" in Installing and Configuring Oracle GoldenGate for Oracle Database and "BATCHSQL".

Syntax

THREAD (thread_ID)
thread_ID

A numerical identifier for the thread that will process this MAP statement. Valid values are 1 through the value that was specified with the MAXTHREADS option of the ADD REPLICAT command that created this group. You can use the INFO REPLICAT command to verify the maximum number of threads allowed for a Replicat group. When specifying thread IDs, the following must be true:

  • The total number of threads specified across all MAP statements of a Replicat group cannot exceed the value of MAXTHREADS.

  • No single thread_ID value in the Replicat group can be higher than the value of MAXTHREADS. For example, if MAXTHREADS is 25, there cannot be a thread_ID of 26 or higher.

If MAXTHREADS was not used, the default maximum number of threads is 25.

Examples

The following examples show some ways to use the THREAD option.

Example 1   

In this example, thread 1 processes table cust.

MAP scott.cust, TARGET scott.cust, THREAD (1);
Example 2   

In this example, thread 1 processes all of the tables in the scott schema.

MAP scott.*, TARGET scott.*, THREAD (1);
Example 3   

In this example, the orders table is partitioned among two MAP statements through the use of FILTER (filter_clause) and the @RANGE function. For more information about @RANGE, see "RANGE".

MAP scott.orders, TARGET scott.orders, FILTER (@RANGE (1, 2, OID)), THREAD (1);
MAP scott.orders, TARGET scott.orders, FILTER (@RANGE (2, 2, OID)), THREAD (2);

THREADRANGE (thread_range, column_list)

THREADRANGE is valid for MAP. This option is valid when Replicat is in coordinated mode.

Use THREADRANGE to specify that the workload of the target table is to be partitioned evenly among a range of Replicat threads, based on the value of a specified column or columns. For example, if the partitioning is based on the value of a column named ID, and the THREADRANGE value is 1-3, then thread 1 processes rows with ID values from 1 through 10, thread 2 processes rows with ID values from 11 through 20, and thread 3 processes rows with ID values from 21 through 30. The partitioning may not be as absolutely even as shown in the preceding example, depending on the initial calculation of the workload, but it is coordinated so that same row is always processed by the same thread. Each specified thread handles filtering, manipulation, error handling, delivery to the target, and other work for its range of rows.

Partitioning a table across a range of threads may improve apply performance for very large tables or tables that frequently incur long-running transactions or heavy volume, but can be used in other cases, as well. You can process more than one table through the same range of threads.

A wildcarded TARGET clause can be used when THREADRANGE is used if the optional column list is omitted. When using a column list, use separate explicit MAP statements for each table that is using the same thread range.

To process a MAP statement with one specific thread, see THREAD (thread_ID). THREAD and THREADRANGE are mutually exclusive options. Do not use them together in the same MAP statement.

Do not specify tables that have referential dependencies among one another in a thread range. Use the THREAD option and process all of those tables with the same thread.

Do not use THREADRANGE to partition sequences. If coordination is required, for example when a sequence is part of a SQLEXEC operation, partition the sequence work to one thread with the THREAD option.

The thread with the lowest thread ID always processes barrier transactions if the THREAD or THREADRANGE option is omitted. Additionally, and work that is not explicitly assigned to a thread is processed through this thread. For example, if there are threads with IDs ranging from 1 to 10, barrier and non-assigned transactions are performed by thread 1.

For more information about Replicat modes, see "Deciding Which Apply Method to Use" in Installing and Configuring Oracle GoldenGate for Oracle Database and "BATCHSQL".

Syntax

THREADRANGE (lowID-highID, [column[, column][, ...]])
lowID

The lowest thread identifier of this range. Valid values are 1 through 500.

highID

The highest thread identifier of this range, which must be a higher number than lowID. Valid values are lowID+1 through 500. The number of threads in the range cannot exceed the value that was specified with the MAXTHREADS option of the ADD REPLICAT command. If MAXTHREADS was not used, the default maximum number of threads is 25.

[column[, column][, ...]]

Optional. Specifies one or more unique columns on which to base the row partitioning. To specify multiple columns, use a comma-delimited list, such as col1, col2, col3. When this option is omitted, the partitioning among the threads is based by default on the following columns, in the order of preference shown:

  • Primary key

  • KEYCOLS clause in the same MAP statement

  • All of the columns of the table that are supported by Oracle GoldenGate for use as a key.

Example

The following example divides the orders and order_lines tables between the same two threads, based on the value of the OID column.

MAP scott.orders, TARGET scott.orders, THREADRANGE (1-2, OID);
MAP scott.order_lines, TARGET scott.order_lines, THREADRANGE (1-2, OID);

TOKENS (token_definition)

TOKENS is valid for TABLE.

Use TOKENS to define a user token and associate it with data. Tokens enable you to extract and store data within the user token area of a trail record header. Token data can be retrieved and used in many ways to customize the way that Oracle GoldenGate delivers data. For example, you can use token data in column maps, stored procedures called by SQLEXEC, or macros.

To use the defined token data in target tables, use the @TOKEN column-conversion function in the COLMAP clause of a Replicat MAP statement. The @TOKEN function maps the name of a token to a target column.

Do not use this option for tables being processed in pass-through mode by a data-pump Extract group.

The character set of token data is not converted. The token must be in the character set of the source database for Extract and in the character set of the target database for Replicat.

Do not use this option for source tables that are encoded as EBCDIC on a z/OS system if the target tables are not EBCDIC.

For more information about using tokens, see Administering Oracle GoldenGate for Windows and UNIX.

Syntax

TOKENS (token_name = token_data [, ...])
token_name

A name of your choice for the token. It can be any number of valid characters and is not case-sensitive. Multi-byte names are not supported.

token_data

Any valid character string of up to 2000 bytes. The data can be either a literal that is enclosed within single quotes (or double quotes if NOUSEANSISQLQUOTES is in use) or the result of an Oracle GoldenGate column-conversion function. See "USEANSISQLQUOTES | NOUSEANSISQLQUOTES" for more information.

Example

The following creates tokens named TK-OSUSER, TK-GROUP, and TK-HOST and maps them to token data obtained with the @GETENV function.

TABLE ora.oratest, TOKENS (
TK-OSUSER = @GETENV ('GGENVIRONMENT' , 'OSUSERNAME'),
TK-GROUP = @GETENV ('GGENVIRONMENT' , 'GROUPNAME')
TK-HOST =  @GETENV ('GGENVIRONMENT' , 'HOSTNAME'));

TRIMSPACES | NOTRIMSPACES

TRIMSPACES and NOTRIMSPACES are valid for TABLE and MAP.

Use TRIMSPACES and NOTRIMSPACES at the root level of a parameter file or within a TABLE or MAP statement to control whether or not trailing spaces in a source CHAR column are truncated when applied to a target CHAR or VARCHAR column. The default is TRIMSPACES.

See "TRIMSPACES | NOTRIMSPACES" for syntax and usage.

TRIMVARSPACES | NOTRIMVARSPACES

TRIMVARSPACES and NOTRIMVARSPACES are valid for TABLE and MAP.

Use TRIMVARSPACES and NOTRIMVARSPACES at the root level of a parameter file or within a TABLE or MAP statement to control whether or not trailing spaces in a source VARCHAR column are truncated when applied to a target CHAR or VARCHAR column. The default is NOTRIMVARSPACES.

See "TRIMVARSPACES | NOTRIMVARSPACES" for syntax and usage.

WHERE (clause)

WHERE is valid for TABLE and MAP.

Use WHERE to select records based on a conditional statement. WHERE does not support the following:

  • Columns that have a multi-byte character set or a character set that is incompatible with the character set of the local operating system.

  • The evaluation of the before image of a primary key column in the conditional statement as part of a primary key update operation.

Enclose literals in single quotes. Specify case-sensitive column names as they are stored in the database, and enclose them in double quotes if the database requires quotes to enforce case-sensitivity (such as Oracle).

Getting More Information about Record Filtering

See Administering Oracle GoldenGate for Windows and UNIX for more information about WHERE and other filtering options.

Syntax

WHERE (clause)
clause

Selects records based on a condition, such as:

WHERE (branch = 'NY')

Table 3-35 shows permissible WHERE operators.

Table 3-35 Permissible WHERE Operators

Operator Example

Column names

PRODUCT_AMT
"Product_Amt"

Numeric values

-123, 5500.123

Literal strings enclosed in single quotes

 'AUTO', 'Ca'

Column tests

@NULL, @PRESENT, @ABSENT (column is null, present or absent in the record). These tests are built into Oracle GoldenGate.

Comparison operators

 =, <>, >, <, >=, <=

Conjunctive operators

AND, OR

Grouping parentheses

Use open and close parentheses for logical grouping of multiple elements.

Example

The following WHERE example returns all records when the AMOUNT column is over 10,000 and does not cause a record to be discarded when AMOUNT is absent.

WHERE (amount = @PRESENT AND amount > 10000)