Tuesday, 15 December 2015

Oracle Data Pump in RAC

Oracle Data Pump was introduced in Oracle Database 10g to enable very high-speed transfer of data and metadata between databases. A salient feature of Data Pump is that it can parallelize the export and import jobs for maximum performance.

However, prior to Oracle Database 11g release 2 (11.2), if a Data Pump job is executed in parallel on an instance of an Real Application Cluster (RAC) database, the parallelism is employed only on that instance without utilizing the potentially valuable idle resources of other instances.

From Oracle Database 11.2 onwards, parallelism is no longer confined to a single instance. Rather worker processes can be distributed across multiple Oracle RAC instances to better utilize Oracle RAC resources. The newly-introduced client parameter CLUSTER can be employed to specify whether you want to use the cluster resources. When CLUSTER=Y (default) is specified, Data Pump attempts to use all active Oracle RAC instances.