Hi,
I support your first approach. Because Master data will always reside outside the Infocubes. You can modify master data any time without consulting Cubes. This concept is Extended Star Schema.
This approach will not touch your 5 cubes at all. It is just a matter of adding one more lookup from DSO2 to create new record.
If we populate the data for 2nd time dependent attribute in ‘MD’ from ‘DSO2’, it should ideally append one more record to master data table instead of overwriting existing record as follows – Please correct me if my assumption is wrong here.
I suggest you to add TDA1 and TDA2 as compounded to your master data infoobject to differentiate two records.
Your second approach is not so feasible to manipulate all the cubes which will lead you to activations and other consequences.
Regards,
Suman