Database Reference
In-Depth Information
Figure 9-3.
Processing data from tables to a cube
This concept is similar to a SQL view with one major difference. The SQL views are just named SQL select
statements that combine results from one or more tables but never stores any data. In contrast, cubes hold a copy
of the actual data from one or more tables. Reports created against the cube cannot access the original tables and
will not be slowed by ongoing transactional activity.
WhY are theY CaLLeD CUBeS?
The term
cube
can be misleading, as it implies a cubes structure has three dimensions. SSAS cubes are
more accurately described as multidimensional data structures. The word
cube
is just easier to say.
Here are some key points to help understand the structure of an SSAS cube:
•
A column is a list of values and can be thought of as a single-dimensional array on a
single attribute. Example: an author's last name.
•
A table is a set of columns and can be thought of as a two-dimensional array on a set
of attributes pertaining to a single subject. Example: an author's basic information,
including address and phone number.
•
A cube can be thought of as a multidimensional array on a set of subjects such as a list
of authors, titles, publishers, and stores multiplied by its measurable values. Each subject
represents a dimension of the cube, and each measurable value is cross multiplied to
provide a distinct aggregate value for each combination of attribute and measure. The
product of which is a multidimensional cube. Figure
9-4
illustrates this concept.
