Creating Data Source Views
Because you will be basing
your cube on a data warehouse/data mart star schema you already have
available, you need to further define exactly what you need to have
access to within that data source. Creating a data source view
essentially allows you to look more deeply into the metadata of the
data source and add additional relationships, create things like
calculations, and set logical keys on the metadata of the data source.
You start by right-clicking the Data Source View object in the Solution
Explorer and selecting New Data Source View (or choosing Project, New
Data Source View). This starts the Data Source View Wizard, which you
use to define what view of data to use for the cube. The first dialog
box allows you to select a data source to use as the basis of the data
source view.
Figure 6 shows the data source Comp Sales2008 that you defined earlier. Chose it and click Next.
If you need to limit the data
source to a particular schema within the database, you can click the
Advanced button and specify a schema (or schemas) to be restricted to
and retrieve any foreign key and primary key relationships that may
exist.
If your schema doesn’t include
foreign key specifications, you can use this wizard to try to discover
foreign key relationships, using a few different types of column name
matching. Figure 7
shows an example of using a simple primary key column name matching
technique to identify any foreign key relationships with other tables
in your schema. If you have used some type of common naming convention
on your source tables, you can easily leverage this name-matching
dialog.
You essentially can identify the following:
Matches based on the exact column name match (as compared to the primary key column):
Order.CustomerID (foreign key) → Customer.CustomerID (primary key)
Matches based on the column name being the primary key table name:
Order.Customer → Customer.CustomerID (primary key)
Matches
based on similar column name by comparing the table name concatenated
with its primary key column name and then loosely comparing it to other
column names of other tables:
Order.CustomerID → CustomerID (concatenated to Customer+ID=CustomerID)
Order.Customer ID → CustomerID
Or Order.Customer_ID → CustomerID
In this example, we used
some good naming conventions for columns, so you can simply specify the
first option (match based on exact column match). This is the lead-in
to select the tables (and/or views) you need to be included from your
data source. As you can see in Figure 8, you can choose from any number of objects. You must select the base tables you need in your data source views. These are the CompSalesFactoid, Geo_Dimension, Prod_Dimension, and Time_Dimension
tables. However, you should also click the Add Related Tables button to
add all related tables, based on the matching technique you specified
earlier. We’ve seen a little inconsistency of the wizard not adding all
related tables properly. Please double check the list of tables with
our figures list (Figure 8). This completes the set of tables that comprise the data source views for your cube.
You now complete this wizard by naming the data source views (Comp Sales2008 DSV) and clicking Finish.
When you exit the wizard, you
end up in the designer view in Visual Studio, with a graphical
representation of the data source views that will be the basis of the
cube you are building (see Figure 9). This figure highlights the primary fact table (CompSalesFactoid), the primary dimension tables (Time_Dimension, Prod_Dimension, Geo_Dimension),
and all tables related to these dimensions (that contain the
values/descriptions of the member entries for the hierarchies of the
dimensions).
Now,
because you have fully specified data source views, you can easily
define a cube via the Cube Wizard. Or you can start defining your
cube’s dimensions and then use these dimensions in the Cube Wizard
later. Because you know your source database well, you should go ahead
and create your dimensions and hierarchies first.
