Cisco Cisco Unified IP Interactive Voice Response (IVR) 8.0(1) 管理员指南

The Unified CCX Cluster uses the publisher/subscriber database model for data 
replication across the system. Under normal circumstances, the publisher acts as 
the source of data and the subscriber acts as the target for the data. 

It is recommended not to install SQL Server with Unified CCX to store your 
custom data. You cannot alter SQL Server configuration settings and this might 
impact the functioning of the contact center.

The publisher/subscriber database model enables Unified CCX to provide 
high-availability and failover support. To support this on the database level, the 
data must be available on multiple nodes of the cluster. To have such data 
availability, replication is used for the Agent, Historical, and Repository 
datastore. (The Configuration datastore does not use replication; instead, it uses 
atomic transaction to commit data changes to all active Configuration datastore in 
the cluster.)

The publisher is the main database. All data is written to this database, with the 
other databases (subscribers) synchronizing with the publisher. If the publisher 
fails, then data can be written to the subscriber(s). When the publisher is back 
online, it returns to accepting writes. It also synchronizes with the subscriber(s) 
by performing the following functions:

Adding any files or records that are new.

Deleting any files or records that have been removed.

Updating any files or records that have a later modification time stamp on the 
subscriber database. 

When the publisher is fully synchronized, then all subscribers return to 
synchronizing with the publisher.

You cannot deactivate publisher datastore components. To deactivate a publisher 
datastore component, first change the component to be a subscriber and then 
activate the component (see 

).