Nice article, but would have liked a mention that using the Multi-phase Datapump to transform data will switch the datapump to row-by-row processing. This is something that people should be aware of and can lead to data loads taking much longer than if a normal datapump copy column task is used.

Also, you don't have to use all the phases. In a lot of my packages I use just the PumpComplete phase. The VBScript in this phase uses the DTSTransformPhaseInfo object to capture the number of rows exported to the file. This object also has an ErrorRows property that holds the number of errors encountered.

yes u are correct phillcart.

I just enable all the phases just to give an idea as u said it is not a compulsory thing to do.

As I mentioned above the main difference is if you use the Multi Phse datapump in data transformations. This switches the datapump into row-by-row processing instead of a bulk load.

If you are able to accept the performance hit you can develop a pretty sophisticated data loadng routine that traps data conversion and key violation errors. The rows in error can be redirected to an exception table for later examination.

If you're loading a lot of data then a better approach would be to bulk load the file directly into a staging table that has all the fields defined as varchar. Then you can run validation routines on the data without affecting performance.