Cisco Cisco Unified Contact Center Enterprise 9.0(2) デザインガイド

Runtime server failover relies heavily on the Unified ICM PG's guarantee that it will connect to only one 
peripheral at a time. Therefore, each runtime server knows it is active if and only if it has an active 
connection from the Unified ICM PG. There are, in fact, no heartbeats or other direct communications 
between the two runtime servers, nor are there any heartbeats or other direct communications between 
the idle side PG and the inactive runtime server.

Each PG and its corresponding runtime server, therefore, fail-over in tandem. If the active runtime server 
fails or is manually taken out of service, then it disconnects itself from its PG, and Unified ICM causes 
the standby runtime server's PG to become active. That PG then establishes a connection to the standby 
runtime server, which causes the standby runtime server to become active.

Similarly, if one PG fails over to its standby counterpart, the corresponding runtime servers fail-over as 
well.

The disposition of calls upon failover depends on the cause of failover. On the one hand, it could be a 
software failover, either because the active node is being shut down manually or because of a software 
failure of some kind. On the other hand, it could be a hardware failover due to either a hardware or 
network outage. An operating system level restart also acts like a hardware failover. The following table 
shows the distinction between these two failover modes.

PG A

PG B

Unified Expert
Advisor
Primary

Unified CCE

187481

Unified Expert
Advisor
Backup