Hot Backup

TPO Hot backup or TPO Clusters is a new capability introduced in the release 5.1 which brings forward the reliability of the system regarding TPO failover.

A TPO cluster must have at least 3 nodes and may contain up to 7 nodes. i.e. N+M >= 3 and N+M <= 7. Multiple TPO Clusters can be used by turrets.

As a general principal, Hot backup is designed to rely on a redundant network. If the deployment take place between DCs, redundant links between DCs are expected. The convergence time is a key factor that may lead to delay TPO failover time to adapt to this network convergence time.

In our terminology, a TPO Cluster is basically a group of multiple TPOs running TPO places whatever their type. Each single element of the group is thus a single TPO, and is called a node. A TPO cluster is a stateful N+M redundancy scheme. The cluster contains N active TPO nodes and M passive TPO nodes. Active nodes take in charge user’s defined places while passive nodes are aware of each calls hosted on active nodes. When an active node fails, a passive node will take in charge the active node's job transparently, without cutting the calls. Failover is designed to take place in a short period of time. After the failover, the end user will still be in position to hold/transfer/conference a call that was initially set on a failing node.

•Virtual IP mode is based on the usage of an virtual IP address for the active node

The following tables gives at a glance availability and difference between methods.

	Alcatel OXE	Avaya Aura	Cisco CUCM	Skype For Business 2015
Virtual IP mode	supported	supported	supported	not supported
DNS mode	supported	not supported	not supported	supported
Session Persistency	not supported	supported	supported***	not supported

	Virtual IP mode	DNS mode	Session Persistency
Redundancy across subnet	no	yes	yes
Redundancy of established calls	yes	yes	yes
Redundancy of incoming calls	yes	partial*	yes**
Connection to PBX	udp	udp or tcp	tcp

* if an incoming call arise between TPO1 failure and TPO2 recovery (more or less 3 seconds), incoming call cannot be saved

** incoming call are saved however if remote party hangup between TPO1 failure and TPO2 recovery or after TPO2 recovery then the turret cannot be notified that the call has disappeared, user will try to accept the incoming call and the call will then be cleaned up.