Cisco Cisco Unified IP Interactive Voice Response (IVR) 8.0(1) Administrator's Guide

Cisco CRS provides high availability and automatic failover capability through 
the use of two servers, the active server and the standby server.

You can deploy over Wide Area Network (WAN) expansion servers on which only 
the Monitoring component or on which only the Monitoring and Recording 
components are installed. All other servers in the cluster must be deployed on the 
same LAN.

The active server provides all system services and resources; no services or 
resources are available from the standby server. Both servers will be synchronized 
when administrative changes are made on the active server. 

If the active server fails, there is automatic failover to the standby server.

When an active server failover occurs, the following happens:

Any active calls existing in the system—and those ringing at an agent’s 
phone—are terminated.

Calls already connected to an agent will not be lost. The agent will be 
logged out of the system but the call in progress will continue.

While the system processes the failover, a time window can occur where calls 
might get rejected or aborted; this time window is five seconds. When the 
failover to the standby server is complete, call treatment, routing, and 
queuing will commence on incoming calls. 

In cases of network failure, failover processing might take a longer time, 
so reconvergence can take more than the standard five-second window. 

All agents will be automatically re-logged in, and put in the ‘unavailable’ 
state. Once automatically re-logged in, each agent must manually change 
their state to what they need it to be.