Geo-separated Telephony Control Server failover experiences a time-delay error
When the active Telephony Control Server node goes down and Telephony Application Server users are not yet connected to the hot-standby node, clients will be unable to make calls.
Geographically separated Telephony Control Servers experience time delays during failover; clients cannot connect to calls until all connections have been reinstated.
Note: This technote also applies to 8.5.2 Interim Feature Release 1.
When the active Telephony Control Server node goes down, the CSTA connection gets broken and must be reconnected to the surviving node (which is a hot-standby). After this, all the Telephony Application Server users must be reconnected, which occurs at a rate of approximately 18 users per second. The total recovery time varies, depending upon how many users are configured on the Telephony Application Server.
For example, a system that has 15,000 users configured would take approximately 13.8 minutes to reconnect all users, where some will get service sooner than others, depending on the order in which they get reconnected. Consider that 20% of 15,000 users is 3000, and at 18 connections per second, the first 3000 users would have service after 167 seconds, or 2.8 minutes.
During this time, all users who have not yet been reconnected are provided with basic telephony service in that incoming calls will always be routed to their "call forward dependable" device.
Advanced telephony features will not be available until all clients have been reconnected.
More support for:
IBM Sametime Unified Telephony
Other Server Issues
Software version: 8.5.2, 22.214.171.124
Operating system(s): Linux
Reference #: 1501483
Modified date: 27 February 2012