You can improve performance when loading data into the data target. Keep the following suggestions in mind.
When include duplicates is selected each incoming row is sent to the target table without first checking to see if the key value is found. This generally provides the best loading time. If there is a duplicate key, the database will reject the row with a unique constraint violation. When reject or update duplicates is selected, the keys of each incoming row are matched against the target table. This additional step does slow processing, but allows for matched rows in the target table to be updated. In this case, care should be taken to use as few columns as keys as possible to uniquely identify the row. Note that if no columns are identified as key columns, the entire column is considered to be the key, which slows processing.
Note that incoming rows are not matched against the target table as with insert/update logic, so the number of rows inserted or rejected are not recorded. There are no flow statistics available.
This method loads faster than the default load logic. For most databases, this is the fastest method, when the query can use automatic passthrough (so no intermediate flat file is created).
iWay Software |