How does InfoCube deletion mode affect loading performance?

[Former Member]

Hi all

we have a scenario where it is necessary to regularly delete the cube before loading.

As we have 2 options for deleting: with or without dimension-tables: what is the more performant way in order to avoid extensive sid-ding the next time the infocube is been loaded?

Normally, I'd expect the variant with only the fact-table to be more efficient as dims are already generated..

But what^s about the "unused" data: it will remain in the dimension-tables then for every time?

And apart from that: if the dimensions are not buffered, will the option "only fact table" really be an advantage?

So, where should the decision be based on?

Regards

bivision

[Former Member · Accepted solution]

hi,

As mentioned by you that both the deletion methods have pros and cons.

The final decision for choosing among the two can be narrowed down by answering some questions such as:

• Is the amount of data loaded daily is huge.

• what is the amount of data which will not come in future loads and will consume DIM table space if not deleted.

• How much time is the loading taking for this cube and What is the business expectation for data availabilty in the report.

regards,

Arvind.

[Former Member · Accepted solution]

Hi

thanks for the quick reply.

So if I don't delete the data from the dimensions tables, is it right that the new loading-process will lookup on those dimension tables first in order to retrieve already existing combinations or how does it process?

So may i limit the generation of new SIDs in this way?

Thanks & regards

bivision