Hello,

I want to create a Report_Lines dimension which contains the rows of any of
the financial reports such as Balance Sheet or Income statement.

I know that the max. level of my reports would be less than 5. A parent
child dimension is very tempting with the following structure with
underlying table structure as:


Report_line_id (Integer)
Report_line_Name (Varchar)
Report_Parent_line_id (Integer)
Report_Line_Type (Varchar - permissible values as 'Report
Name'/'Line'/'Account')
Report_Account_id (Integer - if Report_Line_Type is Account then it contains
the account_id of the account included under Report_Line_Id)
Report_Rollup_Operator (Varchar - permissible values as '~'/'+'/'-' )

However, as a number of reports are defined as it would be run for many
companies, I am thinking it to create a regular dimension instead of Parent
Child so that in my Reporting cube, maximum/all the dimensions are of
regular/ragged type instead of parent-child, so that for intermediate level
aggregations would be stored in cube and the report performance would be
better.

Is my approach correct? If yes, any ideas regarding how I can design my
underlying RDBMS table to store my report ine items.

Thanks,

Ajit.