The success of one-way replication between two environments is dependent on the state of the data in the target environment. If the record that the Replicat process is trying to update does not exist in the target database, the replication will certainly fail. If the data set is small, it is easy to verify that the data in the target environment is in the expected state. However, when the volume of the data increases, it gets tougher and the chances of getting issues in the replication increase manifold. When such errors happen, the Replicat process stops and there can be a pile up of pending changes that still need to be applied to the target environment. In order to avoid the pile up of pending transactions, it might be better to track the error, keep a record of erroneous records, and carry on with the replication.

GoldenGate offers a powerful way of managing such errors. In this recipe we...