Scenario 3 - Different approaches to computing aggregates

Question

In this training you learn two approaches to computing aggregates. In one case, you change the filtering dimension on the Query and then add data of the filtering dimension. In another case, you keep the filtering dimension the same as the targeting dimension and then you add data linked to the filtering (targeting) dimension. Both of these approaches work when computing aggregates. How will your results differ based on the approach that you take?

Answer 1

Currency conversion can be done at two stages.
1. Data Modelling stage
2. Reporting stage
Data Modelling Stage:
*Adv: *
Time taken to have complete data available for reporting is only loading time (background job), so query reporting is very fast, as no calculations at reporting level for currency conversion
Disadv:
If you want to convert to one particular target currency, this approach is preferable but if you want to report at different currencies, you cant convert to all those currencies at modelling side as it leads to huge data in the cubes.
And even this approach is not flexible also, if a new target currency comes to picture, you need to modify the code, load again
Problem with exchange rates also comes into picture here, as at the time of loading you might be considering one exchange rate, at reporting side, if you want to look at data with different exchange rate, its not possible.
Reporting Stage:
Adv:
Can report on any target currency and at any exchange rate by defining calculations at query designer level. This process is flexible enough.
Disadv:
Many calculations at query level might lead to performace issues at runtime.