Cisco Cisco IP Contact Center Release 4.6.1 Design Guide

In summary, the Unified CM call processing service is separate from the CTI Manager service, which 
connects to the Unified CM PG via JTAPI. The Unified CM call processing service is responsible for 
registering the IP phones, and its failure does not affect the Unified CM PGs. From a Cisco Unified CCE 
perspective, the PG does not go off-line because the Unified CM server running CTI Manager remains 
operational. Therefore, the PG does not need to fail-over.

When a CTI Manager service fails, the Unified IP IVR JTAPI subsystem shuts down and restarts by 
trying to connect to the secondary CTI Manager service on a backup Unified CM subscriber in the 
cluster. In addition, all voice calls at this Unified IP IVR are dropped. If there is an available secondary 
CTI Manager service on a backup subscriber, the Unified IP IVR logs into this CTI Manager service on 
that subscriber and re-registers all the CTI ports associated with the Unified IP IVR JTAPI user. After 
all the Unified CM devices are successfully registered with the Unified IP IVR JTAPI user, the server 
resumes its Voice Response Unit (VRU) functions and handles new calls. This action does not impact 
the Unified CVP because it does not depend upon the Unified CM CTI Manager service for call control.

Unified IP IVR Release 3.5 provided for cold standby and Release 4.0 provides hot standby redundancy, 
but this configuration is not recommended for use with Unified CCE. These designs make use of a 
redundant server that is not used unless there is a failure of the primary Unified IP IVR server. However, 
during this failover processing, all calls that are in queue or treatment are dropped on the Unified IP IVR 
as part of the failover. A more resilient design would be to deploy a second (or more) Unified IP IVR 
server(s) and have them all active, allowing the Unified CCE to load-balance calls across them 
automatically. As shown in 

, if one of the Unified IP IVR servers should fail, only the calls 

on that server would fail, but the other active servers would remain active and be able to accept new calls 
in the system. 

The Unified ICM is a collection of services and processes running on Unified ICM servers. The failover 
and recovery process for each of these services is unique and requires carefully examination to 
understand the impact to other parts of the Unified CCE solution, including another Unified ICM service.

As stated previously, all redundant Unified ICM servers discussed in this chapter must be located at the 
same site and connected through a local private LAN or crossover cable. You can provide the private 
LAN by installing a second network interface card (NIC) in each server (sides A and B) and connecting 
them with a crossover cable. By doing so, you can eliminate all external network equipment failures.

When the active CTI Manager Service or PG software fails, the PG JTAPI Gateway/PIM detects an 
OUT_OF_SERVICE event and induces a failover to the redundant (duplex) PG. Because the redundant 
PG is logged into the backup Unified CM subscriber CTI Manager Service already, it registers the IP 
phones and configured dialed numbers or CTI route points automatically. This initialization service 
takes place at a rate of about 5 devices per second. The agent desktops show them as being logged out 
or not ready, and a message displays stating that their routing client or peripheral (Unified CM) has gone 
off-line. (This warning can be turned on or off, depending on the administrator's preference.) All agents 
and supervisors lose their desktop third-party call control functionality until the failure recovery is 
complete. The agents and supervisors can recognize this event because call control action buttons on the 
desktop will gray out, and they will not be able to do anything with the desktop. Any existing calls should 
remain active without any impact to the caller.