on 10-28-2012 4:22 PM
Hey,
Some random Attribute View questions I wanted to ask which is considered best practice. Please explain your answers for better understanding:
1) Is it better to:
a) Have as many related tables in one attribute view to make it very organized. I prefer this method as it is better to test data and seems more organized.
or
b) Try to split up Attribute Views if possible into more views and join all of them at the Analytic View level (better for performance because they will run in parallel?)
^Which practice is better (please explain) and is it not a big deal to use either approach?
2) If there is a transaction table (ex. AFVC) that I need, but I don't need any measures from this table, just attributes. Is it better to:
a) Use the transaction table at the Attribute level
or
b) Use the transaction table at the Analytic Level and create a dummy measure?
^The above scenario gets interesting if there is another transaction table which will be used to get measures. If option b is used then there has to be 2 analytic views created.
3) Related questions:
a) What's the difference between 'calculated attribute' and 'calculated column'?
b) What is the benefit of using one over the other?
c) Any performance benefit for using either?
Mic:
I recommend the below modelling approach:
1) Group tables as master data and transction data. The attribute views will be master data tables and transaction data will be typically analytical views.
2) The analytical views in addition will include business logic that needs to be applied for reporting purpose to link transaction data with master data
3) As needed, based on comple requirements - you will have to design calculation views
Hope this helps.
Regards,
Rama
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
It depends on your model.
For example, you can only use attributes from one table in the AT to join with the AN.
It won't let you join the fact table with key attributes coming from different tables in your AT.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
83 | |
23 | |
11 | |
9 | |
8 | |
5 | |
5 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.