Cisco Cisco Computer Telephony Integration OS 8.5 Technical References

Compared to some other protocols, TCP/IP is relatively slow at detecting and recovering from 
communication path failures. If an IP packet is dropped within the network, retransmission does not 
occur until the sender notices a time-out. This time-out period is usually long enough to allow for 
worst-case round-trip delays and network congestion. If you need more rapid error detection, you may 
send an optional HEARTBEAT_REQ message to the CTI Server whenever no messages have been sent 
for the heartbeat interval. Upon receipt of a HEARTBEAT_REQ message, the CTI Server should 
immediately respond with a HEARTBEAT_CONF message. If three heartbeats go unconfirmed, the CTI 
client should declare a session failure and reset the TCP connection.

You must determine the appropriate heartbeat interval for a production environment—it depends on the 
application and the environment. It should represent a reasonable balance between the speed of failure 
detection and the network bandwidth consumed by heartbeat messages and their corresponding 
confirmations. In cases where there are very few CTI clients, such as a CTI Bridge, the minimum 
heartbeat interval of 5 seconds should suffice. Workstation (desktop) clients should use a much larger 
heartbeat interval (at least 90 seconds), since these clients typically number into the hundreds and 
possibly thousands. A Heartbeat Interval of –1 disables heartbeats. The default setting for application 
developers is –1. However, if the TCP/IP time-out period is adequate, or if there is nothing the 
application can do even if it is aware that something is wrong, it may be appropriate to disable heartbeats 
even in a production environment.

 depicts the heartbeat message flow. 

 define the HEARTBEAT_REQ 

and HEARTBEAT_CONF messages, respectively.