Cisco Cisco IP Contact Center Release 4.6.2 Design Guide

A might connect to the Unified IP IVR at site B. This behavior can create challenges in case of 
connectivity issues between side A & B. Cisco does not support using clustering over the WAN with the 
Unified CCE System PG nor with the System Unified CCE deployment models. Only traditional Unified 
CCE models are recommended for clustering over the WAN.

This section describes the behavior of clustering over the WAN for Unified CCE during certain failure 
situations. The stability of the highly available (HA) WAN is extremely critical in this deployment 
model, and failure of the highly available WAN is considered outside the bounds of what would normally 
happen.

For illustrations of the deployment models described in this section, refer to the figures shown previously 
for 

.

Loss of the entire central site is defined as the loss of all communications with a central site, as if the 
site were switched off. This can result from natural disasters, power issues, major connectivity issues, 
and human error, among other things. If a central site retains some but not all connectivity, it is not 
considered a site loss but rather a partial connectivity loss, and this scenario is covered in subsequent 
sections.

When an entire central site has completely lost Unified CCE clustering over the WAN, Unified MAs will 
fail-over properly to the redundant site.   Failover times can range from 1 to 60 seconds for agents. 
Variations are due to agent count, phone registration location, and agent desktop server used.

When using distributed VoiceXML gateways and Unified CVP, the gateways must fail-over from one site 
to another if their primary site is lost. This failover takes approximately 30 seconds, and calls coming 
into the remote gateways during those 30 seconds will be lost.

If the private connection between Unified ICM Central Controller sides A and B should fail, one Unified 
ICM Router will go out-of-service and the other Unified ICM Router will then be running in simplex 
mode until the link is reinstated. This situation will not cause any call loss or failure. 

If the private connection between PG side A and PG side B should fail, the non-active PG will go 
out-of-service, causing the active PG to run in simplex mode until the link is reinstated. This situation 
will not cause any call loss or failure. 

When using a combined private link, Unified ICM Central Controller and PG private connections will 
be lost if the link is lost. This will cause both components to switch to simplex mode as described above. 
This situation will not cause any call loss or failure. 

If connectivity to one of the central sites is lost from a Unified MA site, all phones and agent desktops 
will immediately switch to the second central site and begin processing calls. Failover typically takes 
between 1 and 60 seconds.