Cisco Systems 1.4 Manual De Usuario

During server installation, you must specify the same Cisco CallManager publisher for all of the 
Personal Assistant and speech-recognition servers in the same Personal Assistant cluster.

Configure the Personal Assistant servers.

During server configuration, you can configure the Personal Assistant servers either to balance the 
call load among themselves or to support failover. If you use failover, you need more servers than 
you would otherwise need for a given number of media ports.

Configure the speech-recognition servers.

You must also identify at least one license manager for the speech software. The speech software 
requires that an active, valid license be available at all times for it to work. See the 

 for information about adding speech-recognition servers 

to the server cluster.

See the following sections for detailed discussions on determining the number of Personal Assistant 
servers required, how to use load balancing, and how failover affects your calculations:

The quantity of servers you install should be adequate to support the number of sessions defined in the 
Personal Assistant server cluster (that is, the sum of sessions on all active Personal Assistant servers in 
the cluster).

The number of Personal Assistant servers and speech-recognition servers that are required in your 
clusters depends on several factors:

The number of concurrent calls to Personal Assistant that you need to support. For example, a sales 
and marketing organization that is very phone-dependent would probably need more servers than an 
engineering group that uses the phone less frequently.

When considering this, you should make separate calculations of the number of simultaneous 
sessions with Personal Assistant and the number of simultaneous sessions with the 
speech-recognition server. For example:

Call-interception sessions—the number of users who will set up rules to enable 
Personal Assistant to intercept calls for them.

Speech-recognition sessions—the number of users who will access voice mail and dial other 
users by name.

The server model that you use. A more powerful server can support more concurrent calls than a less 
powerful server.

Whether you run the Personal Assistant servers and speech-recognition servers on the same system. 
Running both servers on a single system reduces the number of concurrent calls the server can 
support.