Cisco Cisco Unified Intelligent Contact Management Software Leaflet

Page of 428
 
12-6
Cisco Unified Contact Center Enterprise 7.5 SRND
Chapter 12      Bandwidth Provisioning and QoS Considerations
Unified CCE Network Architecture Overview
There are several parameters associated with heartbeats. In general, you should leave these parameters 
set to their system default values. Some of these values are specified when a connection is established, 
while others can be specified by setting values in the Microsoft Windows 2000 registry. The two values 
of most interest are: 
The amount of time between heartbeats
The number of missed heartbeats (currently hard-coded as 5) that the system uses to determine 
whether a circuit has apparently failed
The default value for the heartbeat interval is 100 milliseconds between the duplexed sides, meaning that 
one side can detect the failure of the circuit or the other side within 500 ms. Prior to Unified ICM 
Release 5.0, the default heartbeat interval between a central site and a peripheral gateway was 400 ms, 
meaning that the circuit failure threshold was 2 seconds in this case.
In Unified ICM Releases 5.0 and 6.0, as a part of the Unified ICM QoS implementation, the UDP 
heartbeat is replaced by a TCP keep-alive message in the public network connecting a Central Controller 
to a Peripheral Gateway. (An exception is that, when a Unified ICM Release 5.0 or 6.0 Central Controller 
talks to a PG that is prior to Release 5.0, the communication automatically reverts to the UDP 
mechanism.) Note that the UDP heartbeat remains unchanged in the private network connecting 
duplexed sites. 
In Unified ICM Release 7.x, a consistent heartbeat or keep-alive mechanism is enforced for both the 
public and private network interface. When QoS is enabled on the network interface, a TCP keep-alive 
message is sent; otherwise UDP heartbeats are retained.
The TCP keep-alive feature, provided in the TCP stack, detects inactivity and in that case causes the 
server/client side to terminate. It operates by sending probe packets (namely, keep-alive packets) across 
a connection after the connection has been idle for a certain period, and the connection is considered 
down if a keep-alive response from the other side is not heard. Microsoft Windows 2000/2003 allows 
you to specify keep-alive parameters on a per-connection basis. For Unified ICM public connections, 
the keep-alive timeout is set to 5
400 ms, meaning that a failure can be detected after 2 seconds, as was 
the case with the UDP heartbeat prior to Release 5.0. 
The reasons for moving to TCP keep-alive with QoS enabled are as follows:
In a converged network, algorithms used by routers to handle network congestion conditions can 
have different effects on TCP and UDP. As a result, delays and congestion experienced by UDP 
heartbeat traffic can have, in some cases, little correspondence to the TCP connections.
The use of UDP heartbeats creates deployment complexities in a firewall environment. The dynamic 
port allocation for heartbeat communications makes it necessary to open a large range of port 
numbers, thus defeating the original purpose of the firewall device.