Failover

Collaboration Application Server

As described in the Availability page, if F5 HA is used for Collaboration Application Server (AS) in active-standby mode, the failover can be achieved as the active AS goes out of service.

Collaboration MCU

If an MCU is taken out of service or fails completely, the users are moved to the next most suitable MCU. This can happen in two ways:

• The client loses connection to the MCU. It contacts the AS again, letting the AS know that it lost connectivity. The AS provides an alternative MCU and checks the faulty MCU server using the same network interface that the client would use.
• The AS itself detects a problem with the MCU using results from a comprehensive set of regularly repeated tests which cover the SIP, Flash, and HTTP services. The tests are run on the MCU and results reported to the AS along with load statistics every few seconds. The MCU is then marked as unavailable until further tests succeed. Any clients currently waiting on that server will be moved to another server.

Currently, for a one-to-one media session, the platform does not provide continuity of the session when a failure occurs at MCU level. When a failure happens during such a session, the session is considered to have failed.

However, in case of conference calls, the client auto-redials and the AS then reallocates to a new MCU. The call disconnects for a moment, but then is immediately re-established.