Cisco IM and Presence Service achieves high availability (HA) through subclusters. A subcluster is a logical group of two nodes within the same geographic cluster. One node acts as the primary for a set of users, and the other acts as the backup. When a node fails, such as the Americas publisher, all users homed on that node automatically fail over to the other node within that same subcluster. The intercluster peer connection to EMEAR is for presence federation and cross-cluster messaging, not for user failover. Therefore, the failover is contained within the Americas cluster and specifically to the subscriber node paired with the publisher in the same logical group (subcluster).

A. Failover does not occur across geographic clusters. The intercluster peer to EMEAR is for federation, not for high availability of user services.
B. High availability failover is strictly contained within a subcluster. Users cannot fail over to a node in a different logical group or subcluster.
C. Failover is not supported between different geographic clusters (Americas to EMEAR). The concept of a "same logical group" does not span separate clusters.

1. Cisco Collaboration System 12.x Solution Reference Network Designs (SRND)

Reference: Chapter: Cisco IM and Presence Service

Section: IM and Presence High Availability.

Content: "The IM and Presence cluster architecture is based on subclusters

with each subcluster containing a pair of IM and Presence nodes for high availability (HA). If one of the nodes in a subcluster fails

the users homed on that node are automatically failed-over to the other node in the subcluster." This confirms failover is to the peer node within the same subcluster.

2. Configuration and Administration Guide for IM and Presence Service

Release 12.5(1)SU1

Reference: Chapter: High Availability on IM and Presence Service

Section: High Availability for IM and Presence Service.

Content: "High availability is achieved by creating logical groups of IM and Presence nodes called subclusters. Each subcluster provides redundancy for the users that are assigned to it. If a node in a subcluster fails

the user services fail over to the other node in the subcluster." This explicitly states failover occurs to the other node in the subcluster.

3. Configuration and Administration Guide for IM and Presence Service

Release 12.5(1)SU1

Reference: Chapter: Intercluster Deployments for IM and Presence Service

Section: Intercluster Peer Connections.

Content: This section describes the function of intercluster peers for routing presence and instant messaging traffic between separate IM and Presence clusters. It makes no mention of providing user failover or high availability

confirming its purpose is federation

not redundancy.