Data Warehouse Maintenance, Evolution and Versioning - Data Warehousing Design and Advanced Engineering Applications - page 181

Figure 3. Architecture of DWT (Adapted from Eder and Wiggisser (2006))

in which these transformations are executed, thus
to minimize the graph.
The architecture of a prototypic implementa-
tion of the COMET Metamodel, called DW T is
shown in Figure 3[REMOVED REF FIELD].
The whole system is based on the backend da-
tabase, which contains the versioned structure
information, the transformation functions and the
cell data. DWT T is independent from any specific
data warehouse system. For each system to be
supported, an import and export interface has to
be implemented. These interfaces are responsible
for reading/writing structure and cell data from/
to the particular system. Structure data read from
an external data warehouse system is compared
to the last version stored in the backend database
(if available) by the difference detection compo-
nent. If there is some external knowledge about
the changes available in computer readable form

it can be imported as well. With the differences
generated the version management component
creates a new structure version and stores it into
the backend database. When reestablishing a
particular structure version from the backend
database, the structure selection component reads
all necessary data from the backend database.
Cell data and transformation data are handled
by data transformation component. The export
interface is responsible for writing the structure
data the transformed cell data into the external
data warehouse system.

OTHER APPROACHES ADDRESSING
THE MAINTENANCE PROBLEM

There are several other approaches dealing with
the data warehouse maintenance problem. Most