Maintenance Manual

Chapter 2 Fault Tolerance
Architecture
2-4
Cisco ICM Enterprise Edition Administrator Guide Release 6.0(0)
Central Controller. The Central Controller Agent is a software process that
manages communications between the Central Controller and nodes in the ICM
system.
At any one time, one of the two communications paths is active and the other is
idle. All communication traffic between the Central Controller and the device is
sent on the active path. If the active path fails for any reason, the second path is
activated and all traffic is switched to the newly active path. The previously active
path becomes the idle path.
The communication protocols use buffering and acknowledgments to ensure that
no messages are lost during the path failure and switch-over. After a
communication path failure, the device periodically attempts to re-establish
communication along the failed path.
A different mechanism is used for the real-time data feed to Admin Workstations.
See Real-Time Distributors, page 2-19, later in this chapter, for more information.
Node Manager
Each ICM component (except the client-only Admin Workstation) includes a
Node Manager process. The Node Manager is in charge of restarting Intelligent
Contact Management processes that have failed.
For example, each Logger and each CallRouter has its own Node Manager. If a
Logger and CallRouter are installed on the same machine, two separate Node
Managers run on that machine. If Loggers for multiple customers run on a single
machine, a separate Node Manager runs for each customer.
When a failure occurs in a single-customer ICM system, the Node Manager may
shut down the machine to initiate a reboot. However, in a network service
provider (NSP) environment when a Logger or CallRouter fails, components for
other customers might still be active on the machine. In such a case, the Node
Manager for an NSP component does not shut down and reboot the machine, and
manual intervention is required to restore the failed component.
If the Node Manager does initiate a reboot, the Node Manager itself restarts when
the machine reboots. The Node Manager then starts the other processes for the
component. On a Distributor Admin Workstation, you can choose whether to have
the Node Manager automatically restart when the computer reboots.
For more information on Node Manager start-up options, see the Cisco ICM
Enterprise Edition Installation Guide.