Building Basic Dimensions and Cubes - Expert Cube Development with SSAS Multidimensional Models - page 34

It will also allow you to show your end users something quickly too, so that they
can ensure it meets their expectations and that they can check over it for problems.
This doesn't mean you shouldn't concentrate on getting what you build at this stage
right first time though, far from it, but it is easier to get the details correct when
you're concentrating on a small subset of the solution you hope to build eventually.
We therefore recommend you to pick the most important fact table in your data
warehouse plus a few of the more straightforward dimensions that join to it, one of
which should be the Time dimension, and open up SQL Server Data Tools ( SSDT ).
In the previous versions of Analysis Services, SQL Server Data
Tools was known as BI Development Studio ( BIDS ). SSDT and
BIDS are just different names for the BI project templates inside
Visual Studio.
Multidimensional and Tabular models
As mentioned in the introduction, this topic only concerns itself with Analysis
Services Multidimensional models. With Analysis Services 2012 , however, there is a
second type of Analysis Services: Tabular models. Analysis Services Tabular models
do much the same thing as Analysis Services Multidimensional models—an end user
would probably not be able to tell if they were querying one type or the other—but
the development experience and the underlying technology of the two types of
models are very different.
The development experience for Analysis Services Tabular models is very similar to
that of Power Pivot: the basic concepts are relational, you load data into tables and
you create relationships between tables; there's a big contrast between this and the
world of dimensions, attributes, measure groups, and cubes that we'll encounter
in this topic. The approach the Tabular model takes makes it very easy to create
simple models, but arguably makes it harder to deal with complex requirements:
for example, the ability to handle many-to-many relationships is built into Analysis
Services Multidimensional, but needs to be coded into measure definitions in
Analysis Services Tabular.
Analysis Services Tabular also uses a different way of storing its data to Analysis
Services Multidimensional. Tabular models store their data in an in-memory, column
store database and this can mean that certain operations (such as distinct counts) are
much faster. That said, for most data volumes, both versions of Analysis Services
perform adequately and any performance differences are likely to be negligible.