Exploring Virtual Cubes
Much as a
"view" can logically combine tables, as well as other views, within a
relational database, a virtual cube is a "logical cube" that
is created through the combination of multiple cubes. We select the dimensions
and measures of the virtual cube from the "pooled" dimensions and
measures that exist within the cubes that we are consolidating; we do not need
to use all dimensions and measures but can select those that provide us the
most ideal "view" of the data that we are attempting to produce from
the disparate data sources. As a logical cube, the virtual cube appears as a
single cube from the perspective of the information consumer.
While virtual cubes
often consist of selected dimensions and measures from multiple cubes, we will
likely encounter situations in our organizations where, rather than combining
the data from multiple OLAP data sources, we wish to restrict the data we
present to information consumers to a subset of the dimensions and measures
within a single cube. This represents yet another scenario where a virtual
cube might provide an excellent option.
Other scenarios exist
that are well addressed by the flexible functionality that virtual cubes
readily offer. These include the use of linked cubes in addition to normal
cubes in their creation.
Strengths of virtual
cubes include their flexibility, portability and ease of creation and
modification. In addition, the storage space required by a virtual cube is
modest: since virtual cubes store only their own definitions, leaving the
storage of the data of the cubes that underlie them strictly to the cubes
within which the data resides, physical storage space is almost negligible. The
minimal storage requirement adds handsomely to the flexibility factor, as it
enables us to create combinations and variations of existing cubes without the
introduction of material overhead into the equation.
The capacity of a virtual
cube to act as an adjunct to an overall security strategy is also a strength.
The physical security provided by the virtual cube lies in the options it
provides to restrict access for selected information consumer groups to
sensitive or other selected information.
As we mentioned before, a
virtual cube can be based upon a linked cube, but we need to keep in mind the
limitations when this is the case, including the fact that the virtual cube
does not support (for the linked cube):
- cell calculations
- custom member formulas
- custom rollup operators
- custom rollup formulas
Creating Virtual Cubes
We will create virtual
cubes for a couple of the main reasons that they are typically created:
to consolidate information that exists in more than one cube, and to create a
subset of information in a cube/combined cubes as a means of controlling its
presentation/accessibility to information consumers. We will point out the
reasons that these options might be valuable to information consumers or
management of the organization as we undertake the steps involved.
Let's get started with
an example that mirrors the primary reason for creating a virtual cube: to
consolidate information. One common reason we might want to do this could be
to allow for comparison between measures that exist in two separate cubes.