Databases Reference
In-Depth Information
Three-stage architecture
Now, let's take a look at the three-stage architecture:
In this architecture, two additional layers are added:
Transformation Layer
and
Transformed QVD Layer
. The role of these two additional layers is to hold all
transformations that need to be performed upon the source data before it can be
integrated into the target data model.
This also suggests that all base QVDs will ideally keep a straight copy of the source
table, which will optimize the extraction process. Then, the Transformation Layer, in
which several base QVDs will be combined to create denormalized QVDs, performs
any required aggregation or segmentation, and adds new calculated fields or
composite keys to prepare the transformed QVDs for a clean and simple load into
the final data model.
