Hi Karthik,
Removing determinants may have impact on existing reports, since this information is used at query planning time. So all the reasons that would require determinants to be present apply.

If you use one level of granularity then you probably can get away with fewer determinants. However you may have scenarios where the data is at a single level of granularity in each query subject, but once you combine them, then it becomes critical to the query planner to understand the granularity with respect to each other.
See: Multiple-fact, Multiple-grain Queries - IBM Documentation

Hi Karthik,
Removing determinants may have impact on existing reports, since this information is used at query planning time. So all the reasons that would require determinants to be present apply.

If you use one level of granularity then you probably can get away with fewer determinants. However you may have scenarios where the data is at a single level of granularity in each query subject, but once you combine them, then it becomes critical to the query planner to understand the granularity with respect to each other.
See: Multiple-fact, Multiple-grain Queries - IBM Documentation

The other item being the ability to avoid using a set operation (DISTINCT/GROUP BY) to produce distinct rows, if the columns participate in a unique determinant.

------------------------------
NIGEL CAMPBELL
------------------------------


Original Message:
Sent: Tue April 19, 2022 09:44 AM
From: HENK CAZEMIER
Subject: Framework Manager - Determinants

Hi Karthik,
Removing determinants may have impact on existing reports, since this information is used at query planning time. So all the reasons that would require determinants to be present apply.

If you use one level of granularity then you probably can get away with fewer determinants. However you may have scenarios where the data is at a single level of granularity in each query subject, but once you combine them, then it becomes critical to the query planner to understand the granularity with respect to each other.
See: Multiple-fact, Multiple-grain Queries - IBM Documentation

Hi,

The determinant you're showing above isn't really achieving anything useful for the query engine. If you were to remove it, you'd get the same results. The time determinants become important is when your data is denormalized within a query subject. Commonly you'll find this occurring in star schema dimensions - where the different "levels" of descriptive information have been pulled together in one place. In this situation, you can often get fact query subjects at different granular levels joining to the one dimension query subject. This is where determinants in the dimension query subject play a critical role in identifying the granular groupings the query engine needs to be aware of.

For any query subjects with a single unique determinant like the one you're showing here, this is usually superfluous. If you were to merge it with other query subjects at different grains, then you'd need to evaluate whether determinants are required in the new merged query subject.

Cheers!

MF.