COGNOiSe.com - The IBM Cognos Community

IBM Cognos Analytics Platform => Cognos Analytics => Framework Manager => Topic started by: andigrill on 23 Sep 2022 07:01:05 AM

Title: Dimensional Functions in Framework Manager
Post by: andigrill on 23 Sep 2022 07:01:05 AM
Hello,

Until recently we used Cognos Transformer for creating Cubes. Because it's still 32-Bit and doesn't work too well with huge data we are starting modelling cubes in Framework Manager.
Datasource are Oracle Tables.

However, there are still some features of Transformer we are not sure how to implement in FM:
In time dimensons we want to have the same 'Current Month','Last month' Month before last month' and so on like transformer automatically creates.
Of course it's possible to have a table with the needed Dates and filter them for alternating hierarchies.
But how can you use the cognos Dimensional functions?
I tried
tail ([Dimensional Layer].[Date Dimensional].[Date].[Year]) and testing it delivers the correct value.
When I publish the package and open the Cube in Analysis Studio and expand the Hierarchy I get the message:
GEN-ERR-0016 : The 'tail' Function is Not Supported in the Context in Which It is Used

Where is the error?
Title: Re: Dimensional Functions in Framework Manager
Post by: cognostechie on 24 Sep 2022 05:36:54 PM
Did you know that 'huge data' doesn't work in Framework either and that's exactly IBM introduced Dynamic Cubes.
Title: Re: Dimensional Functions in Framework Manager
Post by: andigrill on 26 Sep 2022 07:08:19 AM
We tried DynCubes too and the Performance is pretty similar.

Title: Re: Dimensional Functions in Framework Manager
Post by: cognostechie on 26 Sep 2022 03:28:20 PM
What's the definition of 'huge data' ? How many rows from each transaction/fact table?

It might be possible that the cubes are not built properly or the database is not efficient. I have had cubes with 90 million+ rows and worked with acceptable performance.