DD Questions
3 posters
Page 1 of 1
DD Questions
Dear all,
I have doubts on the usage of Degenerate Dimension (DD) and need expert's advices.
For example, ticket number, if there are no attributes for the ticket, then I could make it a DD, right?
However my thinking is if in the overall model, there are more than 1 fact tables storing the ticket number (as DD) and somehow these fact tables needed to be joined together by the ticket number, then would it be an inefficient query (fact to fact join without an intermediate dimension)? Could I just make a dimension having fields TICKET_SK and TICKET_NO?
Thanks a lot!
I have doubts on the usage of Degenerate Dimension (DD) and need expert's advices.
For example, ticket number, if there are no attributes for the ticket, then I could make it a DD, right?
However my thinking is if in the overall model, there are more than 1 fact tables storing the ticket number (as DD) and somehow these fact tables needed to be joined together by the ticket number, then would it be an inefficient query (fact to fact join without an intermediate dimension)? Could I just make a dimension having fields TICKET_SK and TICKET_NO?
Thanks a lot!
kinsun- Posts : 6
Join date : 2012-03-22
Re: DD Questions
I don't get the point if an 'intermediate dimension'. When we design a model, we draw lines between a fact and a dimension. All because another fact table uses the same dimension, it does not mean you actually need to join to the dimension table when all you want to do is join between fact table keys. The lines are simply displaying the nature of the relationship. It doesn't mean you have to join that way.
The other thing, as a matter of form, the two facts involved should be aggregated to the same grain before joining. This avoids any possibility of a many-to-many join between the two facts.
The other thing, as a matter of form, the two facts involved should be aggregated to the same grain before joining. This avoids any possibility of a many-to-many join between the two facts.
Re: DD Questions
Thanks Ngalemmo for the clear explanation.
I have this question because I was taught in the past, direct fact table to fact table join (F-F) would result in slow query performance. It is always good to have a dimension in the between (F-D-F). This is the reason that I raised this question.
Anyways thanks for the new insight given to me!
I have this question because I was taught in the past, direct fact table to fact table join (F-F) would result in slow query performance. It is always good to have a dimension in the between (F-D-F). This is the reason that I raised this question.
Anyways thanks for the new insight given to me!
kinsun- Posts : 6
Join date : 2012-03-22
Re: DD Questions
You need to create a query which brings data from first fact and have one row for each ticket number and then do the same with other fact and then join theses to resultset.
Dimension modellings soul is in not having facts repeatation due to dimensions that is the reason at a point in time we have only one active record for a particular thing in the dimensions and for the same reason nobody suggest to join fact with a fact.
But if still you want to do it make sure you take care of first 2 lines above.
Dimension modellings soul is in not having facts repeatation due to dimensions that is the reason at a point in time we have only one active record for a particular thing in the dimensions and for the same reason nobody suggest to join fact with a fact.
But if still you want to do it make sure you take care of first 2 lines above.
Vishy- Posts : 75
Join date : 2012-02-28
Location : India
Similar topics
» Specific Questions (Basic Questions)
» NDS to DDS Design Questions
» Dimensions with effective and end dates - best design and best way to join to fact tables?
» Bus Matrix Questions
» Data Vault v's Dimensional Model
» NDS to DDS Design Questions
» Dimensions with effective and end dates - best design and best way to join to fact tables?
» Bus Matrix Questions
» Data Vault v's Dimensional Model
Page 1 of 1
Permissions in this forum:
You cannot reply to topics in this forum
|
|