Cisco Cisco IP Contact Center Release 4.6.2 Dépliant

Server count — The number of servers that are required to manage the parent/child model is usually 
higher due to the increased number of software components (additional Gateway PGs, additional 
Central Controller for each child, and so forth).

The Unified CCE Gateway PG, Unified CM cluster, Unified IP IVR, and Unified System CCE 
should (if possible) be co-located at the contact center site.

The communication link from the parent Unified ICM Central Controller to the Unified CCE 
Gateway PG must be sized properly and provisioned for bandwidth and QoS. (For details, refer to 
the chapter on 

Gatekeeper-based or RSVP agent-based call admission control could be used to reroute calls 
between sites over the PSTN when WAN bandwidth is not available. It is best to ensure that 
adequate WAN bandwidth exists between sites for the maximum amount of calling that can occur.

If the communication link between the Unified CCE Gateway PG and the parent Unified ICM 
Central Controller is lost, then all contact center routing for calls at that site is put under control of 
the local Unified System CCE. Unified CVP-controlled ingress voice gateways would have 
survivability TCL scripts to redirect inbound calls to local Unified CM CTI route points, and the 
local Unified IP IVR would be used to handle local queue and treatment during the WAN outage. 
This is a major feature of the parent/child model to provide complete local survivability for the call 
center. For more information, see the chapter on 

While two intercluster call legs for the same call will not cause unnecessary RTP streams, two 
separate call signaling control paths will remain intact between the two clusters (producing logical 
hairpinning and reducing the number of intercluster trunks by two).

Latency between parent Unified ICM Central Controllers and remote Unified CCE Gateway PGs 
cannot exceed 200 ms one way (400 ms round-trip).

This deployment model is the same as the previous model, except that Unified CVP is used instead of 
Unified IP IVR for call treatment and queuing. In this model, voice gateways with PSTN trunks 
terminate into each site. Just as in the centralized call processing model with distributed voice gateways, 
it might be desirable to limit the routing of calls to agents within the site where the call arrived (to reduce 
WAN bandwidth). Call treatment and queuing can also be achieved at the site where the call arrived, 
further reducing the WAN bandwidth needs. 

 illustrates this model using traditional Unified 

CCE deployment.