Database Reference
In-Depth Information
Figure 1. Schema and instance changes in the example company
From the view of structural changes, these
reorganizations do not cause any problems. But
what about the data? Table 1 shows the cell data
for the cube in the years 2005, 2006 and 2007, i.e.
for the Versions 1, 2 and 3 respectively. A possible
chart resulting from this data without considering
changes may look like shown in Figure 2. Without
knowing about the changes, interpretation of this
chart is not possible.
user defined attributes. Also the hierarchical rela-
tions between different categories may be subject
to change. On the instance level, the system must
provide operations for the
Insertion, Deletion
and
Change
of dimension members, as well as
operations for changing the hierarchical relations
between the dimension members.
Recording and managing structure changes is
only one aspect of the maintenance problem. The
cell data contained in the data warehouse depends
on the structure. Thus, modifications of the structure
may lead to inconsistencies in the cell data. So it
may have to be adjusted to be consistent again. Such
adjustments could range from simple reaggrega-
tion to complex data transformations, because, for
instance, the unit of a fact has changed. For the
example given above, a data adjustment may be the
recalculation for the brand Mercedes, representing
that Model G is now sold under this brand.
Levels of the Maintenance Problem
As defined above, the cube's structure consists of
schema and instances. A data warehouse mainte-
nance system must be able to keep track of changes
in both of them. Thus, on the schema level it must
provide operations for
Insertion, Deletion
and
Change
of dimension and categories. Category
changes are, for instance, addition or deletion of
Search WWH ::
Custom Search