Cisco Systems 1.4 Creating a Personal Assistant Server Cluster With Failover

1-15

Cisco Personal Assistant 1.4 Installation and Administration Guide

OL-4590-03

Chapter 1 Planning for Personal Assistant

Creating Server Clusters

If you are not using failover servers, and if a Personal Assistant server becomes disabled, the interceptor

port route points registered with that particular server are unavailable. Because the remaining servers

were not configured as failover servers, these interceptor ports cannot re-register with these servers.

Instead, Personal Assistant cannot intercept calls for these extensions. Thus, all rule-based call routing

for the affected users will be unavailable and all calls will ring directly through to the applicable

extensions.

Creating a Personal Assistant Server Cluster With Failover

If load balancing, as explained in the “Setting Up Personal Assistant Server Load Balancing” section on

page 1-13, does not provide you with sufficient redundancy, you can configure failover servers in the

Personal Assistant cluster.

If you are using failover and a Personal Assistant server becomes disabled, the failover server takes

control of the media ports and interceptor ports that were configured on the disabled server. For example,

if you configured 15 media ports on the disabled server, the failover server would add 15 media ports to

its configuration. Thus, if you use failover servers, you must have twice as many servers for a given

number of media ports as would be required if you were not using failover servers.

Although the failover server takes on the media and interceptor ports of the disabled server, it cannot

take over active calls. Any active calls on the disabled server are dropped. However, if Personal Assistant

has completed its role in the call process (for example, it had transferred a call based on call-routing

rules), the call remains in progress.

In addition to taking over the disabled server ports, the failover server registers itself with

Cisco CallManager as the disabled server CTI route point.

When the disabled server becomes active again, it asks the failover server to return its ports. The failover

server returns the ports as they become available; no active calls are dropped. When the reactivated

server regains all media ports, it reregisters itself as the CTI route point with Cisco CallManager.

There are two main techniques for setting up failover servers:

• Using Active Personal Assistant Servers for Failover, page 1-15

• Using Spare Personal Assistant Servers for Failover, page 1-16

When you use an active Personal Assistant server as a failover server, the server works as a regular

Personal Assistant server managing calls with users. The server is not idle.

However, if the primary server becomes disabled, the failover server must be able to handle the media

and interceptor ports of the disabled server, as well as its own. Thus, you must have sufficient capacity

on the failover server to accommodate the ports defined on the disabled server.

For example, if you are using two MCS-7835-H1-IPC1 Personal Assistant servers, each server supports

a maximum of 36 media ports (see Table 1-2). If you use the servers as failover servers for each other,

you must reduce the media ports on each server to no more than 18. So, if server A goes down, server B

will take over the 18 ports of server A, and server B will temporarily run with 36 ports (its original 18

plus the 18 of server A).