Full vs. Delta Load

[Former Member]

Hi BI-Experts!

This is a rather general question.

Is there a recommendation when to use Full or Delta Load?

I know the functional differences but do SAP or some technical reasons say that each of them may only be used in special cases?

For example Delta Load can only be used with SAP source systems?

Best regards,

Philipp

[Former Member · Accepted solution]

Hi Philipp,

This is a pretty open-ended question. I will give you a few examples of when I use each ....

Delta's:

1.) If the record set is large and changes a lot daily. If you have a large initial load, let's say 2Million records and you get about 200k records per month, then you want to use a delta mechanism. There are just too many records to reload daily.

2.) If the record set is wide (many fields instead of a few), it will take less load time to run a delta than a full load.

3.) Datasource restrictions....some business content datasources don't have full loads and vice versa (some don't have deltas).

4.) Do my loads take too long? If so, I will convert full loads to delta's to shave off time here and there.

Full

1.) I gauge this mainly based on load time. If it takes 5 minutes to load 100k records, then it's going to take longer to figure out what delta's were missed (if there was a problem) than it would to just reload the dataset.

2.) Do my requirements dictate that I need to drop and reload the dataset? The best example of this would be the current year's forecast. Some companies revise the forecast and don't want to keep any previous revisions.

3.) Some data loads load a snapshot of a specific date. So, if I wanted to take a snapshot of inventory in a DSO/ODS on Monday, I would Datamart it into a Cube doing a full load for Monday's date.

Hope it helps!

Thanks,

Brian