Cisco Cisco Customer Voice Portal Downloads

High Availability H.323-Based Call Flow

This section describes the relationship among Unified CVP solution components from the
perspective of an H.323-based call flow. We begin with a hypothetical comprehensive model
incoming call, and trace its effect on each component with a particular view toward how that
component finds the next component downstream. In the process, we will be describing the
system’s high availability architecture.

Please refer to the following diagram.

Figure 8: H.323-Based Unified CVP Solution

Although not shown in the diagram, a Media Server can be connected to the VXML Gateway.
Contact Server Switches (CSSs) can be placed between the VXML Gateway and the VXML
Server, between the VXML Gateway and the ASR/TTS Server, between the VXML Gateway
and the Media Server.

PSTN should be
contracted to fail over to

Multiple Ingress
Gateways can be

New call from PSTN to Ingress Gateway

other TDM connections

deployed, each with its

when the selected
connection is down.

own array of TDM
connections.

H.323 Services that are
not ready to take calls are

If dial peer session target
is RAS (Registration,

New Call from Ingress Gateway to Call Server (H.323 Service)

not registered with the

Admission, and Status),

gatekeeper and will not be

then one of up to two

given calls. If no H.323

gatekeepers are consulted

Services are available, the

for the IP address of an

gatekeeper returns no

H.323 Service that is

destination, and the

known to be available and

Ingress Gateway will

ready to take calls.

reject the call.

Gatekeeper can distribute

Alternatively, this

calls to targets in round

situation can be caught by
the Survivability script.

robin or preference order,
and can also track

Planning Guide for Cisco Unified Customer Voice Portal Cisco Unified Customer Voice Portal 4.1(1)