Difference between attribute and default attribute hierarchy? - ssas

Question

Often attributes have their own attribute hierarchy by default. Not a user hierarchy. Just to make sure I'm clear on my terminology:
Attribute hierarchy:
[Dim City].[State Name]
Attribute:
[Dim City].[State Name].[State Name]
Attribute member:
[Dim City].[State Name].[State Name].&[Kansas]
From the standpoint of writing MDX queries, most of the time using either in a cross join gives desired results. Occasionally when I'm doing more complex queries, I hit situations where using the [Dim State].[State Name] in a cross join will give me combinations of values which aren't related.
[Dim City].[State Name]
* [Dim City.[City Name]
Usually using one or the other produces identical results, sometimes not. In more complex queries sometimes I get combinations of unrelated members like (Atlanta, Kansas), and then I change it to below and it works correctly:
[Dim City].[State Name].[State Name]
* [Dim City.[City Name].[City Name]
My main questions is to not figure out why those specific queries may not work correctly, but really what the difference is between using on or the other.
It seems like both produce a set of the same members.
Both seem to have individual members directly within them:
[Dim City].[State Name].[State Name].&[Kansas]
[Dim City].[State Name].&[Kansas]
I understand user hierarchies, but have never really figured out what the functional difference is with these default attribute hierarchies.

Answer 1

Usually, each attribute hierarchy exists of two levels: A top level that contains the 'All' member and the bottom level that contains the attribute elements.
The overall structure of the parts of a dimension in an MDX expression are like the following:
[Dimension name].[Hierarchy name].[Hierarchy level name].[Member Name]
[Dimension name].[Hierarchy name].[Hierarchy level name].&[Member Key]
Your first example
[Dim City].[State Name]
* [Dim City.[City Name]
is a shortcut for
[Dim City].[State Name].Members
* [Dim City.[City Name].Members
So, [Dim City].[State Name].Members would be resolved to {[Dim City].[State Name].[All], [Dim City].[State Name].[Alabama], [Dim City].[State Name].[Alaska], [Dim City].[State Name].[Arizona], ...} because you call the Members function on the hierarchy and not the level.
In your second example you include the level name and thus you'll get only the list of state names without the 'All' member.
If you'll get an invalid combination of state and city names in some queries you should check your attribute relationships. The city name attribute should point to the state name attribute, but you should do this only, if your key values of your attributes follow this relationship. E.g. if you have cities with the same name in several states, every city needs its own key.
BTW: You should use & instead of $ in your MDX queries, but I assume that to be a typo.