Building Scalable Analysis Services Applications - Microsoft SQL Server 2008 Analysis Services - page 460

one of your servers with information about the state of the market. You can create linked
measure groups to enable your customers to link to your data. Your customers can use
data in your measure group to analyze their sales against the market figures you provide.

NOTE
If you change source measure groups often, you might find it useful to create separate
cubes, one per measure group, and create a single cube that joins all the measure
groups you need as linked measure groups. This way you can avoid a situation where
you change one measure group and Analysis Services drops the caches for the whole
cube. If the data in one of the measure groups changes, the caches of cube with the
linked measure groups won't be affected.

Remote Partitions

You can use the remote-partitions feature of Analysis Services to build a distributed back-
end solution. Doing so enables you to scale a volume of stored data and increase the speed
of processing and query performance.

In a remote-partitions scenario, you work with two kinds of servers: a single master server
and one or more remote servers. The master server serves as a primary storage of meta-
data, provides access to the data for client applications, and performs all the management
operations. The remote servers contain partitions with data—remote partitions.

To create a remote partition, you create a database on a remote server and set up a data
source object that points to the master server. The remote Analysis Services uses this
connection to get information, such as dimension data, from the master server. On the
master server, you create a data source object that points to the remote server. Then you
create a partition using a DDL similar to the DDL you use to create a regular partition. The
difference is that you specify a RemoteDatasourceID property that points to the data
source you just created.

As an example, let's assume that you have two servers: the master server (localhost) and
the remote server (localhost\slave). You create a remote partition, Inventory Fact 1997 ,
in the FoodMart 2008 database. See Figure 25.5 for a diagram of this example.

To set up the remote partition Inventory Fact 1997 on the remote server, follow these
steps:

1. Create a new database, Remote DB, on the remote server.
2. Create a new data source object on the remote server that uses the Microsoft OLE DB
Provider for Analysis Services 10.0 to point to the FoodMart 2008 database on the
master server. Listing 25.4 is the DDL of the Create command.

LISTING 25.4

DDL to Create a Data Source Pointing to the Master Server

<Create xmlns=”http://schemas.microsoft.com/analysisservices/2003/engine”>
<ParentObject>