Extraction, Transformation, and Loading - Data Warehouse Systems: Design and Implementation - page 316

Fig. 8.39 Load of the Continent ( a )andthe Country ( b ) dimension tables
Figure 8.39 a shows the transformation that loads the Continent dimension
table. With respect to the corresponding data flow in Fig. 8.30 b, we can
see that the conversion task is not required in Kettle. On the other hand,
the transformation that loads the Country dimension table in Fig. 8.39 b
differs from the corresponding transformation in Integration Services in
Fig. 8.30 c, since in Kettle, we can find the ContinentName associated with
agiven Country using an XPath expression in a similar way as was done in
the conceptual design given in Sect. 8.3 . On the other hand, in Integration
Services, a merge join is needed for this task. After this, a lookup over the
database is performed to get the ContinentKey , and finally the Country table
is loaded.
Fig. 8.40 Load of the City dimension table
Figure 8.40 shows the transformation that loads the City dimension table.
It significantly differs from the corresponding transformation in Fig. 8.32 in
several aspects. First, it is not possible to cascade lookup steps in Kettle as
it is done with lookup tasks in Integration Services since in the latter the
lookup task allows one to split the records into alternative flows depending
on whether a lookup value is found or not, while there is no simple way to do
this in Kettle. Thus, the cascade lookups must be implemented as a collection
of parallel flows. For the same reason, in the Kettle implementation, we do