How to apply very large batch jobs or transactions in a replicating environment (4298532)

How to apply very large batch jobs or transactions in a replicating environment.

Normally it is preferred that the extremely large transactions or batch jobs be applied independently on source and target. This will prevent the replication from getting overloaded. The two most convenient ways to do it are as follows:

A. Ignore the actions of the user for the duration of the transaction:

1. Configure the Capture parameter SP_OCT_DENIED_USERID on source to ignore the actions of a specific user for the duration of the transaction:

sp_ctrl>set param SP_OCT_DENIED_USERID  <Oracle userid for the specific user>

1. Let that user apply the transactions on source which can later be applied on target as well since Shareplex will  ignore that transaction. 
2. Once this is done, unset the parameter SP_OCT_DENIED_USERID:

sp_ctrl>reset param SP_OCT_DENIED_USERID

 For more details on the nuances of setting the above parameter, refer to Knowledgebase Article # 55971 as well as Shareplex Reference Guide Chapter titled “Shareplex Parameters Section titled “Capture Parameters”.

B. Invoke the procedure SHAREPLEX_IGNORE_TRANS before the start of the transactions so that the transaction is ignored by Shareplex.  Please refer to Shareplex Administrator Guide section titled “Skip large maintenance transactions “.

C. There is yet another way to ignore a particular type of transaction by configuring the Transformation routine in Shareplex. However, this will not be efficient when ignoring large transaction and one would be better off by not ignoring the transaction and letting Shareplex handle it. Thus it would defeat our purpose.  Moreover it is cumbersome to set up so if this goal of ignoring the transactions is only a one-off event, Transformation is not the desired way of doing it. For details on Transformation, please refer to Shareplex Administrator Guide Chapter titled “Transform data “.