The Agent PG communicates with the Unified Communications Manager cluster through the CTI Manager. An Agent PG can control agent phones and CTI route points anywhere in the cluster. The Agent PG registers with the CTI Manager on a Unified Communications Manager subscriber in the cluster. The CTI Manager accepts all JTAPI requests from the PG for the cluster. When the PG requests a phone or route point on another subscriber, the CTI Manager forwards the request to the other subscriber using the CallManager SDL links.
This section uses Agent PG to describe any PG that includes the Unified Communications Manager PIM. The Agent PG can be a Generic PG or a Unified Communications Manager PG. For example, the following figure uses Generic PGs to connect to the subscribers. Those Generic PGs are acting as Agent PGs.
A fault-tolerant design deploys Unified CCE Agent PGs in a redundant configuration, because a PG only connects to the cluster through a single CTI Manager. If that CTI Manager fails, the PG cannot communicate with the cluster. A redundant PG provides a second pathway through a different CTI Manager on a different subscriber in the cluster.
The minimum design for a high-availability cluster is one publisher and two subscribers. If the primary subscriber fails, the devices rehome to the secondary subscriber and not to the publisher for the cluster.
Figure 8. High Availability Design for Unified Communications Manager Cluster
The redundant PG servers keep in synchronization through a private network that is isolated from the public network. If the two servers run on different physical machines at the same site, you can create the private network by connecting an Ethernet Cross-Over Cable between their private-network NICs. If the two PG servers are geographically distributed, use a separate WAN connection for the private network. To avoid a single point of failure in the network, do not use the same circuits or network gear as for the public network.
Within the Agent PG, the JTAPI Gateway and Agent PG PIM manage the connectivity to the cluster. The JTAPI Gateway handles the JTAPI socket connection protocol and messaging between the PIM and the CTI Manager. The PIM manages the interface between Unified CCE, the JTAPI Gateway, and the cluster. It requests specific objects to monitor and handle route requests from the cluster. The PG starts the JTAPI Gateway and PIM automatically as node-managed processes. The PG monitors the processes and automatically restarts them if they fail.
During PG installation, download the JTAPI Gateway from the cluster to ensure compatibility. Whenever you upgrade either the PG or Unified Communications Manager, remove and reinstall the JTAPI Gateway.
The JTAPI services from both redundant Agent PGs sign in to the CTI Manager after initialization. Agent PG Side A signs in to the primary CTI Manager; Agent PG Side B signs in to the secondary CTI Manager. Only one PG in each pair actively registers and monitors phones and CTI route points. The redundant PG runs in hot-standby mode. The redundant PG signs into the secondary CTI Manager only to initialize the interface and make it available for a failover. This arrangement significantly decreases the time for the failover.
During system start, the PG that first connects to the Unified CCE Call Router server and requests configuration information is the active PG. The Call Router ensures that the PG side that has the best connection becomes active. The nominal designations of “Side A” and “Side B” do not affect which PG becomes active. During a PG failover caused by a private link failure, a weighting mechanism chooses which PG is active to minimize the impact on the contact center.
The PIM startup process registers all CTI route points first, which is done at a rate of five route points per second. For systems with large numbers of CTI route points, it can take several minutes before the system allows any agents to sign in. You can reduce this time by distributing the devices over multiple PIM interfaces to the cluster.
If calls arrive at the CTI Route Points before the PIM is operational, the calls fail unless you set up the route points with a recovery number. Place the recovery number in their Call Forward on Unregistered or Call Forward on Failure setting. For example, you can set the recovery numbers to the Cisco Unity voicemail system for the Auto Attendant.