Regarding constraint conditions for the DI Studio table loader transformation

[EinarRoed]

Regarding the Table Loader transformation in DI Studio. I have some difficulty understanding when the constraint conditions should be set to Before Load, and when they should be set to After Load. Are any of you able to explain this to me, or give me a good rule-of-thumb about it? I know the job runs much quicker when the constraint conditions are set After Load, but I don't want to risk anything by always doing it that way. So basically, when should I set constraint conditions Before Load, and why?

Thanks for your time.

[LinusH · Accepted solution]

Hard to give a rule of thumb, it all depends on your data, data cleansing, data model etc.

How do you intend to use constraints?

If you use the ETL process to cleanse your data in most of the aspects that you can think, any constraint will be used as extra security, and not often triggered during load. Then it could be wise to apply constraints after load.

What you risk is by applying after load, is that non consistent data may be loaded/updated in the target table, and you need to figure out how to roll those inserts/updates out.

Another aspect is data volume vs the load window, is performance an issue?

If not, you could have constraints applied permanently.