Microsoft Office Tutorials and References
In Depth Information
Now that you know what normalization is, you still need to learn how to handle normalized
tables and how and when you should denormalize them. In the rest of this chapter, we
provide several examples for you to follow, taking practical samples from the AdventureWorks
database. To give you an advance briefing of these techniques, you always have two options
for performing denormalization:
■ You can load normalized tables into PowerPivot and then use the RELATED function to
provide denormalized columns in the original table.
■ You can denormalize the database and load it in a denormalized format into PowerPivot,
taking advantage of the fact that PowerPivot automatically normalizes the data, thus
optimizing its storage.
Nevertheless, before we dive into a complete treatment of how to denormalize tables, we
want to spend some words on another very important difference between the physical and
the logical data models so that you get a complete scenario.
Another big difference between technical and logical structure of the database is in the
treatment of empty values. From a purely technical point of view, a missing value is exactly
what it is: missing. For example, take a look at the report in Figure 4-16, which is built using
the same sales tables that we have shown you up to now and that you can find in the
companion workbook CH04-03-EmptyValues.xlsx.
FIguRE 4-16 Empty values, in the database, are empty.