Building Scalable Analysis Services Applications - Microsoft SQL Server 2008 Analysis Services - page 463

processes the partition data. (For more information, see Chapter 21, “Dimension and
Partition Processing.”)
At the end of the remote-partition processing, the remote server contains a multidimen-
sional cube with linked dimensions, a regular measure group with partitions that contain
data. Then, the remote server sends metadata information about the partition—such as
the partition slice, partition size, indexes built for the partition, and so on—to the master
server. This information enables the master server to avoid performing a round trip to the
remote server for query execution and to make decisions about whether the requested
data is stored on that remote partition.
The processing of multiple remote partitions is coordinated by the master server. It sends
the Process commands to all the remote servers and coordinates the processing transac-
tions between itself and each of them. A single processing transaction involves several
remote servers. After the transaction is committed, a new version of the data from all
remote servers becomes available to the users simultaneously, across all the servers. (For
information about querying remote partitions, see Chapter 30, “Architecture of Query
Execution—Retrieving Data from Storage.”)
Using Business Intelligence Development Studio to
Create Linked Dimensions
It's simple to define a new linked dimension. We'll use our sample FoodMart 2008 project.
In BI Dev Studio, change the name of the project we've been working with from FoodMart
2008 to FoodMart 2008 Linked. Right-click the project node and select Properties from the
resulting menu. On the deployment page of the Project's Properties dialog box, change
Target Database to FoodMart 2008 Linked. Save the changes and deploy the solution.
Now your Analysis Server contains two databases: FoodMart 2008 and FoodMart 2008
Linked. We're going to link the Product dimension from FoodMart 2008 as a
ProductStandard dimension into the FoodMart 2008 Linked database:
1. Right-click the Dimensions node in the Solution Explorer and select New Linked
Dimension from the resulting menu. The Linked Object Wizard opens.
2. In the wizard, advance to the Select Data Source page, where you can create a new
data source that points to the database you want to link your dimension to.
3. For our example, create a new data source that points to the FoodMart 2008 sample
database you've deployed to the local server.
4. Advance to the Select Objects page and select the Product dimension.
5. Click Finish on the Completing the Wizard page. A new linked dimension appears in
your database. It's named Product 1 , so change the name to ProductStandard .
6. Open the Warehouse and Sales cube in the cube editor and add the
ProductStandard dimension to Warehouse and Sales cube. Use the Dimension
Usage tab to define the relationship between the ProductStandard dimension and
the Warehouse measure group. Figure 25.6 shows the Define Relationship dialog box
that appears on top of the cube editor.