Cisco Cisco Customer Voice Portal 8.0(1) Design Guide

Page of 223
 
4-20
Cisco Unified Customer Voice Portal (CVP) 8.x Solution Reference Network Design (SRND)
OL-15989-06
Chapter 4      Designing Unified CVP for High Availability
Unified CVP H.323 Service
Configuration
Unified CVP H.323 configuration for high availability is performed primarily on the ingress gateways, 
but it is also necessary to configure the H.323 Service to register to the gatekeeper.
Configuring High Availability for New Calls
The gatekeeper knows which Call Servers are in service or out of service. It is therefore important to let 
a gatekeeper route incoming calls to a Call Server. By default, Unified CVP H.323 Services register to 
the gatekeeper with a technology prefix (tech-prefix) of 2#. The Unified CVP H.323 Service must 
register with a tech-prefix, and it is not possible to configure the H.323 Service without a tech-prefix.
A technology prefix is a way for the gatekeeper to categorize registering endpoints by functionality. In 
general, no additional configuration is needed on the gatekeeper for incoming calls. The H.323 Service 
registers to the gatekeeper with 2#, and the originating gateway prepends a 2# to the incoming Dialed 
Number Identification Service (DNIS) digits. The gatekeeper automatically knows how to match the 
gateway request to an available Call Server. On the gatekeeper, the command show gatekeeper 
gw-type-prefix
 displays the route plan that the gatekeeper uses to route calls.
On the originating gateways, define the dial-peer for the Call Servers as follows:
dial-peer voice 11111 voip
 session target ras 
 tech-prefix 2# 
The command session target ras instructs the gateway to send the call to its gatekeeper. The command 
tech-prefix 2# instructs the gateway to prepend 2# to the DNIS number when sending the call to the 
gatekeeper.
Configuring High Availability for Calls in Progress
In the event that a Call Server fails with calls in progress, it is possible to salvage all calls if certain 
gateway configuration steps have been taken. A Call Server can fail in one of the following ways:
  •
The server can crash.
  •
The process can crash.
  •
The process can hang.
  •
There can be a network outage.
The configuration discussed in this section protects against all of these situations. However, the 
following two situations cannot be protected against:
  •
Someone stops the process with calls in progress. This situation occurs when a system administrator 
forgets to put the Call Server out-of-service first to allow calls in progress to finish before stopping 
the process.
  •
The Call Server exceeds the recommended call rate. Although there is a throttle for the absolute 
number of calls allowed in the Call Server, there is no throttle for call rate. In general, exceeding 
5 calls per second (cps) for an extended period of time causes the Call Server to have erratic and 
unpredictable behavior. This situation can be prevented by proper sizing of the Unified CVP system.
For call survivability, configure the originating gateways as described in the latest version of the 
Configuration and Administration Guide for Cisco Unified Customer Voice Portal (CVP), available at