Cube Analysis

[Former Member]

I am analysing an info cube and have observed the following:

1. If I click on the cube and take the manage option on the requests tab I see 6 requests (I have taken the date from back to 1900)

2. If in SE16 I go and view the number of entries in the package dimension I have 15 entries.

3. If I take the SIDs for Request ids I can see all values in the SID table for Request ids - 6 SIDs tie up with the 6 requests in manage.

Does this indicate that the cube has been deleted in the past where say 9 times the FACT table data was only deleted and not the Dimension data and if not what is this implying? Thanks

[Former Member]

I think you have the right idea.

Actually, not deleting the dimension table is an option for performance increasing assuming that you did not change the character combinations in the dimensions.

( not creating DIM ID's for the same combination of values ) However, when you make a new request by loading your data, it will get a different ID. Which means, this rule can not be applied to the package dimension at all ( excluding the case of reconstruction from PSA) .

Derya

[Former Member]

Derya, thanks for this. In terms of performance gain, I think we are both talikng about load times.

I am actually trying to dig deeper into the ratio size between FACT and dimension tables and understand why dimensions can be fat - I am aware of the obvious, but would like to understand whether redundant data exists as part of the dimensions (the only case I can see this occurring is deletion of FACT table only) and if so how much this bears a part to the size of the dimension