Deployment Strategies - Microsoft SQL Server 2008 Analysis Services - page 810

After it receives a Synchronize command that contains this <Location> section, the Target
Master server connects to the Target Slave1 server and instructs it to synchronize itself
with Source Slave1. If several slave servers are involved, the Target Master coordinates the
synchronize operation with all the slave servers. Only after all the slaves have finished the
synchronization operation does the Target Server commit the transaction.

Coordination between several servers is a complicated task. You must take care of the
following details when you synchronize a database that has remote partitions:

. Before you synchronize, make sure that all the target slave servers are set up

correctly to contain the slave databases and that the MasterDataSourceID property is
set. For more information about how to set up slave servers, see Chapter 25,
“Building Scalable Analysis Services Applications.”

. Use the default locations for your remote partitions; don't use custom locations.

. If the source master configuration changes so that it no longer keeps partitions on

one of the slave servers, after synchronization you have to delete the slave database
that the target master server no longer uses.

. The target slave server service account needs to have administrative privileges for the

source slave server with which it is synchronizing.

Synchronization and Failover Clusters

The synchronization operation is agnostic as to whether either the source or the target
server is installed and operating on a failover cluster. If the failover operation occurs
during synchronization, Analysis Services automatically rolls back the unfinished synchro-
nization transaction.

You can restart the Synchronize command when the failover operation is complete.