Adding new characteristics to a compressed cube

[Former Member]

Dear Experts

I need to add about 5 new characteristics to an existing inventory cube that is compressed daily.

Currently, this cube has lots of data and 9 + 3 dimensions in total.

The max dimensions allowed is 16 , so I can use another 4 more.

Logically, I would group the 5 into 3 new dimensions.

These are data fields of material document numbers from mseg and mkpf.

These fields are smbln and smblp. Records with these fields populated are few compared to the total number of material documents.

Should I create new dimension to hold these or insert these 2 into existing dimension where the material doc number is ?

Which is the safer or better peformance or better approach?

Here is what I think :

1. If I create 3 new dimensions, I still have 1 left for future use. But what happens when all 16 is used up and later I need 1 more?

    Is performance going to get worse when more dimensions are created? I think so but not sure.

2. If  I insert these into dimensions more suitable for them, such that no new dimension need to be used, how does this affect performance?

Is this risky to data integrity?  We are ok with existing data in BI cube to be blank for these new fields and going forward, these will be populated by extractor.

Kindly please could you advise me the better approach?

Thank you for reading.

Best regards

Pascal

[former_member199691 · Accepted solution]

while creatig dimensions, you need to keep in mind that ration of entries in a dimension tabble should not be more than 10-15 % of data in fact table. Based on this parameter, you should design your dimesions. It is better to have more dimensions with few characteristics in each dimension than fewer dimensions with more characteristics in each dimension.