PowerExchange CDC Publisher can perform source table-level filtering. However, CDC Publisher does not perform source column-level filtering or row-level filtering. All source columns or fields are included in every message. For each change record or row received, the CDC Publisher generates a message.

The PowerExchange CDC Publisher is certified with a single source system and a single target messaging system (a one-to-one topology). The source tables or objects must be all of the same source type and registered under a single source instance. Toplogies that have a single source and multiple target messaging systems (a fan-out topology) or that have multiple sources and a single target messaging system (a fan-in topology) are not currently supported. For more information, contact Informatica Global Customer Support or your product specialist.

The version of PowerExchange that is installed on the PowerExchange CDC Publisher system must match the PowerExchange version that is installed on other systems in the PowerExchange change capture environment.

By default, the CDC Publisher expects that the PowerExchange Logger for Linux, UNIX, and Windows is configured to extract both before-images and after-images of data. If the PowerExchange Logger is configured to extract only after images, the before images do not appear in the Avro messages. In this case, set the following properties when you configure the CDC Publisher configuration files:

Extract.pwxUpdateImageOption=AI
Formatter.avroIncludeBeforeImage=false. The setting of false disables the generation of the Avro fields for before images.

For data sources that were registered in PowerExchange based on a COBOL copybook, the CDC Publisher uses the low-level fields but not the group that contains the fields. To handle this situation, you can use a PowerExchange expression or let the consumer application determine the method to use.

For Oracle sources, if you specify ORACLE_UNHANDLED_NUMASCHAR=Y in the DBMOVER configuration file that the PowerExchange Navigator uses, PowerExchange converts NUMBER columns that have a precision greater than 28 or an undefined length and FLOAT columns that have a precision greater than 15 significant digits to VARCHAR(172) strings when creating capture registrations and extraction maps. The CDC Publisher does not convert the VARCHAR(172) columns back to numeric values.

Avro represents all character strings in UTF-8 character encoding.

If DDL changes are made to source tables, the CDC Publisher ignores the DDL changes and retains the existing message structure until you take action to update the Avro schema. For more information, see Handling Changes to Source Tables and Extraction Maps.