Databases Reference
In-Depth Information
The Attribute dimension
There is a dimension type known as an
Attribute
dimension type. The
Attribute
dimension is not a full blown standard dimension and in fact must be associated
with a standard dimension. The
Attribute
dimension allows you to analyze your
data with even finer granularity by adding data characteristics (for example, color)
to your
Total Vehicle
dimension.
If it is not an
Attribute
dimension type then it is
some form of
Standard
dimension.
Dense and Sparse dimensions
As we described before, the data stored in an Essbase database is stored in a manner
conceptually similar to that of a Rubik's Cube. Because of this structure, the folks at
Oracle have devised a method of defining your outline dimensions and their data
density to Essbase. Therefore, your database can be more efficient in the way it loads,
stores, calculates, and retrieves your data.
The distinction between
Dense
and
Sparse
dimensions allows Essbase to efficiently
handle large amounts of data that is not evenly spread across the data blocks. It
is this efficiency that allows Essbase to offer the slice and dice style of data access,
while still maintaining high performance for fast data processing and retrievals.
Correctly set
Dense
and
Sparse
dimension attributes have a
dramatic effect on overall memory usage and system performance!
• In a
Sparse
dimension, the data does not exist for the better
part of the member combinations. For example, all vehicles
may not be sold in all of the countries
• A
Dense
dimension has a very high percentage of available
data cells filled with data.
First of all, you will need to consider whether a database dimension is
Dense
or
Sparse
in an Application|Database that has been created using the BSO. All
dimensions in an ASO database are
Sparse
by default. When you have an Essbase
database created using the BSO, Essbase allows you to describe your dimensions as
either
Dense
or
Sparse
with the following understanding. Some data cubes, created
roughly by dimension, are not fully populated with data. In fact, for the most part,
the typical Essbase multidimensional database's cube always has a relatively
Sparse
concentration of data at the overall database level. It would be an extreme rarity to
find your data densely populating every available cell with a stored value.
