Configure Extract for a Downstream Deployment

A downstream Oracle GoldenGate deployment allows you to offload the processing of the redo logs to an alternative database referred to as a mining database. The mining database does not need to contain any data from the source database, but must be the same database version or higher.

Note:

The source and downstream database servers must have the same endianness and processor types. For example, downstream Extract will work with source Linux (x86) to a downstream Windows (x86) server, but it does not work if the source server is HP-UX but the downstream server is AIX, or the source server is HP PA-RISC but the downstream server is HP-Itanium.
A downstream mining database accepts archived logs and online redo logs from a source database. This implies that a downstream Extract can exist in two modes:

  • Real-Time Mining mode: In Real-time mining mode, the redo records are processed from the online or standby redo logs immediately, so, there is no delay in processing the database changes.

  • Archive-Log-Only mode: In archive log mining mode, the redo records are not processed until they are archived into a redo log file.

In a typical downstream deployment, the source database, where changes are made to the data, will ship it's redo logs or archive logs to a mining database where the Extract process will mine the redo logs and write out the trail files. In the following diagram, CDBNORTH is the source database and CDBSOUTH is the target database. There is a mining database between the source and target with the database logmining server that transfers redo logs from the downstream Extract to the target.

Workflow between source multitenant container database and a downstream database with logmining server.

An alternative option is to use a cascading approach. If there is an Active Data Guard standby, you can have the standby ship the redo logs to the mining database instead of shipping the logs from the primary. This option should be used if there is limited network bandwidth and the standby database is in the same region or data center as the mining database.

This approach uses an ADG configured in a cascaded mode to transport redo logs to a downstream mining database, which is then referred by a downstream Extract, reducing the overhead on the source database, as shown in the following diagram.

Note:

The Extract process for Oracle cannot connect to a Data Guard standby or Active Data Guard standby to capture data.

Downstream Database connected to Active Data Guard (ADG)

For configuration details about setting up a cascaded downstream environment, see Enable Downstream Extract to Work with ADG.