- Integrated Capture requires Oracle RDBMS release 18.104.22.168 and later.
- Integrated Capture is available in two modes: Local Deployment and Downstream Deployment
Local Deployment means that the source database and the logmining database are the same (that is, running in the same instance), which must be at least release 22.214.171.124
Downstream Deployment means that source database and the logmining database are different databases. The logmining server is created in the downstream instance.
The Redo Logs of the source database are continuously shipped to the downstream instance.
Integrated Capture: Downstream Deployment
- A downstream mining server offloads the capture and transformation process overhead from the production source database thus improving replication performance.
- Downstream deployment is the only option for Integrated Capture from Oracle RDBMS instances that earlier than 126.96.36.199.
- If it is used with a downstream mining server, the release of the source database must be at least 10.2.0.0.
- Downstream deployment offloads database capture and transformation, but it requires log shipping and other configuration changes.
- It is important to note that the Oracle Database on the downstream machine is used to hold minimal data or metadata that is specific to the Oracle Internal Log processing module.
- The standby Redo Logs (or archive Redo Logs) are not applied to the downstream database, but are processed by the Oracle Internal Log processing module and made available as Logical Change Records (LCRs) to the consumer thread of the Integrated Extract process.
- The later is responsible for writing the processed data into the Oracle GoldenGate trail file.
There are two types of Downstream Deployments: Real-Time and Deferred Downstream Deployment.
- Real-Time Downstream Deployment
- In this deployment configuration, Redo Logs are shipped continuously to the downstream machine as “standby Redo Logs”.
- Deferred Downstream Deployment
- When replication SLAs enable a longer lag time between the source and target databases, and when synchronization can be deferred, archive redo files are shipped at regular intervals to the downstream mining database.