Database Reference
In-Depth Information
Figure 8. An example of a) ISA representation and b) schema with c) instances of a generalized hier-
archy
with Occupation type in Figure 8b). Therefore, the
implementation of this kind of hierarchy currently
requires a significant programming effort.
Eilenburg do not have members assigned for the
StateProvince level.
In order to represent non-covering hierarchy
at the logical level, a solution similar to the one
used for unbalanced hierarchies is applied, i.e.,
null values or placeholders are included in the
missing members. Then, applying the mapping
rules leads to a snowflake or star schema, e.g., a
flat DimGeography table in Figure 2, with corre-
sponding attributes in addition to the DimReseller
table, i.e., mixing snowflake and star structures.
SSAS represents a non-covering hierarchy
as shown in Figure 10a. When displaying the
instances, MS provides the same four options as
described for unbalanced hierarchies. However,
for non-covering hierarchies, the third and fourth
options should be applied, since, for our example
in Figure 9b, the parent member with null value
has two children. Unfortunately, SSAS does not
handle this hierarchy well, since the missing level
is always shown as can be seen in Figure 10b.
Non-Covering or Ragged Hierarchies
A non-covering or ragged hierarchy is a spe-
cial case of a generalized hierarchy. It contains
multiple exclusive paths sharing at least the leaf
level; however, alternative paths are obtained by
skipping one or several intermediate levels of
other paths.
The schema in Figure 3 includes a Location
non-covering hierarchy that is also shown in Figure
9a; this hierarchy is composed of the Reseller ,
City , StateProvince , and CountryRegion levels.
However, as can be seen by the straight right-
hand line and the cardinalities, some countries
do not have division into states. Figure 9b shows
some hypothetical instances that we use for this
hierarchy 6 . Notice that the cities of Berlin and
Search WWH ::




Custom Search