Database Reference
In-Depth Information
DISTINCTCOUNT(Table[Column])
. In many cases, distinct count measures in tabu-
lar models can be much more effective than in the multidimensional world.
In multidimensional models, there are a lot of special considerations you need to
take in order to get distinct count to perform well. For more information about this,
please see the following whitepaper:
http://www.microsoft.com/en-us/download/de-
Linked objects
In Analysis Services' multidimensional cubes, you have the concept of linked ob-
jects. They allow you to include dimensions, or measure groups from other data-
basestoacubewithoutduplicatingthedata.Thiscanbeusefulifyouwanttoinclude
objects that could be important to end users analyzing your cube without wasting
disk space. This concept is missing in the tabular mode, where everything is stored
compressed in the memory of the server.
Many-to-many relationships
To many the most crucial feature that is missing from the tabular cubes, is the ability
to create many-to-many relationships between tables. This is something that, in
many cases, pushes the project towards using multidimensional modeling instead.
However, even though it is not officially supported, there is a possibility of creating
a model containing many-to-many relations through DAX scripts. There are sever-
al blogs on the web that describe how to do it. One of them is the following:
ht-
Parent-child hierarchies
Parent-childhierarchiesaredimensionsthatcontainhierarchiesthatreferencethem-
selves through a self-join; this is a feature that is missing from tabular models. In
the multidimensional model, you have it as a native dimension type, and it can be
used to describe objects such as an organizational structure. However, in the tabu-
lar world, you have the ability to create a parent-child hierarchy through DAX code.
However, the code gets very complex as can be seen in the following blog post:
ht-
Search WWH ::
Custom Search