Creating Analysis Services Cubes - SQL Server Analysis Services 2012 Cube Development - page 91

Relationship
Type Explanation and examples

Regular This relationship type implies that the dimension is directly related

to the measure group through one of its attributes, normally the key
attribute. Once you specify the relationship type, you must specify
the dimension column and the corresponding measure group
column (the names of these two columns do not have to match).
You must also specify the granularity. In most cases you use dimen-
sion's key attribute to relate it to the measure group. For example,
the Promotion dimension is directly related to Internet sales through
the Promotion Key. It is possible to relate a dimension through a
non-key attribute; if you do so, you must ensure that attribute rela-
tionships are defined from the relating attribute to all the other attrib-
utes within the dimension. Recall that all non-key dimension attrib-
utes are directly related to the key attribute, and that is how SSAS
ensures that data is aggregated correctly. For example, budgeting
measure groups could be related to the date dimension on month,
quarter, or year attributes instead of the key attribute, which is nor-
mally set to individual dates.

Fact Fact dimensions are built upon the fact table and therefore have a

one-to-one relationship with fact records. Fact tables often contain
millions or billions of records, therefore, fact-level dimensions
should not be exposed for ad hoc queries. Instead, they can be
used for reporting such attributes as an invoice number or an order
revision number. Some literature refers to fact-level dimensions as
degenerate dimensions. For example, the Internet sales order de-
tails dimension has a fact relationship with the Internet sales meas-
ure group. This dimension is exposed using Internet details drill-
through action. You will learn about actions in Chapter 4 , Extending
and Customizing Cubes .